Computer Vision and Pattern Recognition (cs.CV)

Abstract: Backdoor attack aims to compromise a model, which returns an adversary-wanted output when a specific trigger pattern appears yet behaves normally for clean inputs. Current backdoor attacks require changing pixels of clean images, which results in poor stealthiness of attacks and increases the difficulty of the physical implementation. This paper proposes a novel physical invisible backdoor based on camera imaging without changing nature image pixels. Specifically, a compromised model returns a target label for images taken by a particular camera, while it returns correct results for other images. To implement and evaluate the proposed backdoor, we take shots of different objects from multi-angles using multiple smartphones to build a new dataset of 21,500 images. Conventional backdoor attacks work ineffectively with some classical models, such as ResNet18, over the above-mentioned dataset. Therefore, we propose a three-step training strategy to mount the backdoor attack. First, we design and train a camera identification model with the phone IDs to extract the camera fingerprint feature. Subsequently, we elaborate a special network architecture, which is easily compromised by our backdoor attack, by leveraging the attributes of the CFA interpolation algorithm and combining it with the feature extraction block in the camera identification model. Finally, we transfer the backdoor from the elaborated special network architecture to the classical architecture model via teacher-student distillation learning. Since the trigger of our method is related to the specific phone, our attack works effectively in the physical world. Experiment results demonstrate the feasibility of our proposed approach and robustness against various backdoor defenses.

Abstract: This work breaks through the Base-New Tradeoff (BNT)dilemma in prompt tuning, i.e., the better the tuned model generalizes to the base (or target) task, the worse it generalizes to new tasks, and vice versa. Specifically, through an in-depth analysis of the learned features of the base and new tasks, we observe that the BNT stems from a channel bias issue, i.e., the vast majority of feature channels are occupied by base-specific knowledge, resulting in the collapse of taskshared knowledge important to new tasks. To address this, we propose the Decoupled Prompt Tuning (DePT) framework, which decouples base-specific knowledge from feature channels into an isolated feature space during prompt tuning, so as to maximally preserve task-shared knowledge in the original feature space for achieving better zero-shot generalization on new tasks. Importantly, our DePT is orthogonal to existing prompt tuning methods, hence it can improve all of them. Extensive experiments on 11 datasets show the strong flexibility and effectiveness of DePT. Our code and pretrained models are available at https://github.com/Koorye/DePT.

Abstract: With the vigorous development of the urban construction industry, engineering deformation or changes often occur during the construction process. To combat this phenomenon, it is necessary to detect changes in order to detect construction loopholes in time, ensure the integrity of the project and reduce labor costs. Or the inconvenience and injuriousness of the road. In the study of change detection in 3D point clouds, researchers have published various research methods on 3D point clouds. Directly based on but mostly based ontraditional threshold distance methods (C2C, M3C2, M3C2-EP), and some are to convert 3D point clouds into DSM, which loses a lot of original information. Although deep learning is used in remote sensing methods, in terms of change detection of 3D point clouds, it is more converted into two-dimensional patches, and neural networks are rarely applied directly. We prefer that the network is given at the level of pixels or points. Variety. Therefore, in this article, our network builds a network for 3D point cloud change detection, and proposes a new module Cross transformer suitable for change detection. Simultaneously simulate tunneling data for change detection, and do test experiments with our network.

Abstract: Visual localization is the task of estimating a 6-DoF camera pose of a query image within a provided 3D reference map. Thanks to recent advances in various 3D sensors, 3D point clouds are becoming a more accurate and affordable option for building the reference map, but research to match the points of 3D point clouds with pixels in 2D images for visual localization remains challenging. Existing approaches that jointly learn 2D-3D feature matching suffer from low inliers due to representational differences between the two modalities, and the methods that bypass this problem into classification have an issue of poor refinement. In this work, we propose EP2P-Loc, a novel large-scale visual localization method that mitigates such appearance discrepancy and enables end-to-end training for pose estimation. To increase the number of inliers, we propose a simple algorithm to remove invisible 3D points in the image, and find all 2D-3D correspondences without keypoint detection. To reduce memory usage and search complexity, we take a coarse-to-fine approach where we extract patch-level features from 2D images, then perform 2D patch classification on each 3D point, and obtain the exact corresponding 2D pixel coordinates through positional encoding. Finally, for the first time in this task, we employ a differentiable PnP for end-to-end training. In the experiments on newly curated large-scale indoor and outdoor benchmarks based on 2D-3D-S and KITTI, we show that our method achieves the state-of-the-art performance compared to existing visual localization and image-to-point cloud registration methods.

Abstract: Talking Face Generation (TFG) aims to reconstruct facial movements to achieve high natural lip movements from audio and facial features that are under potential connections. Existing TFG methods have made significant advancements to produce natural and realistic images. However, most work rarely takes visual quality into consideration. It is challenging to ensure lip synchronization while avoiding visual quality degradation in cross-modal generation methods. To address this issue, we propose a universal High-Definition Teeth Restoration Network, dubbed HDTR-Net, for arbitrary TFG methods. HDTR-Net can enhance teeth regions at an extremely fast speed while maintaining synchronization, and temporal consistency. In particular, we propose a Fine-Grained Feature Fusion (FGFF) module to effectively capture fine texture feature information around teeth and surrounding regions, and use these features to fine-grain the feature map to enhance the clarity of teeth. Extensive experiments show that our method can be adapted to arbitrary TFG methods without suffering from lip synchronization and frame coherence. Another advantage of HDTR-Net is its real-time generation ability. Also under the condition of high-definition restoration of talking face video synthesis, its inference speed is $300\%$ faster than the current state-of-the-art face restoration based on super-resolution.

Abstract: A recent trend in deep learning algorithms has been towards training large scale models, having high parameter count and trained on big dataset. However, robustness of such large scale models towards real-world settings is still a less-explored topic. In this work, we first benchmark the performance of these models under different perturbations and datasets thereby representing real-world shifts, and highlight their degrading performance under these shifts. We then discuss on how complete model fine-tuning based existing robustification schemes might not be a scalable option given very large scale networks and can also lead them to forget some of the desired characterstics. Finally, we propose a simple and cost-effective method to solve this problem, inspired by knowledge transfer literature. It involves robustifying smaller models, at a lower computation cost, and then use them as teachers to tune a fraction of these large scale networks, reducing the overall computational overhead. We evaluate our proposed method under various vision perturbations including ImageNet-C,R,S,A datasets and also for transfer learning, zero-shot evaluation setups on different datasets. Benchmark results show that our method is able to induce robustness to these large scale models efficiently, requiring significantly lower time and also preserves the transfer learning, zero-shot properties of the original model which none of the existing methods are able to achieve.

Abstract: Generating realistic talking faces is a complex and widely discussed task with numerous applications. In this paper, we present DiffTalker, a novel model designed to generate lifelike talking faces through audio and landmark co-driving. DiffTalker addresses the challenges associated with directly applying diffusion models to audio control, which are traditionally trained on text-image pairs. DiffTalker consists of two agent networks: a transformer-based landmarks completion network for geometric accuracy and a diffusion-based face generation network for texture details. Landmarks play a pivotal role in establishing a seamless connection between the audio and image domains, facilitating the incorporation of knowledge from pre-trained diffusion models. This innovative approach efficiently produces articulate-speaking faces. Experimental results showcase DiffTalker's superior performance in producing clear and geometrically accurate talking faces, all without the need for additional alignment between audio and image features.

Abstract: Despite the remarkable success of deep learning systems over the last decade, a key difference still remains between neural network and human decision-making: As humans, we cannot only form a decision on the spot, but also ponder, revisiting an initial guess from different angles, distilling relevant information, arriving at a better decision. Here, we propose RecycleNet, a latent feature recycling method, instilling the pondering capability for neural networks to refine initial decisions over a number of recycling steps, where outputs are fed back into earlier network layers in an iterative fashion. This approach makes minimal assumptions about the neural network architecture and thus can be implemented in a wide variety of contexts. Using medical image segmentation as the evaluation environment, we show that latent feature recycling enables the network to iteratively refine initial predictions even beyond the iterations seen during training, converging towards an improved decision. We evaluate this across a variety of segmentation benchmarks and show consistent improvements even compared with top-performing segmentation methods. This allows trading increased computation time for improved performance, which can be beneficial, especially for safety-critical applications.

Abstract: This dataset represents almost all the harmful diseases for rice in Bangladesh. This dataset consists of 1106 image of five harmful diseases called Brown Spot, Leaf Scaled, Rice Blast, Rice Turngo, Steath Blight in two different background variation named field background picture and white background picture. Two different background variation helps the dataset to perform more accurately so that the user can use this data for field use as well as white background for decision making. The data is collected from rice field of Dhaka Division. This dataset can use for rice leaf diseases classification, diseases detection using Computer Vision and Pattern Recognition for different rice leaf disease.

Abstract: Remote sensing images are essential for many earth science applications, but their quality can be degraded due to limitations in sensor technology and complex imaging environments. To address this, various remote sensing image deblurring methods have been developed to restore sharp, high-quality images from degraded observational data. However, most traditional model-based deblurring methods usually require predefined hand-craft prior assumptions, which are difficult to handle in complex applications, and most deep learning-based deblurring methods are designed as a black box, lacking transparency and interpretability. In this work, we propose a novel blind deblurring learning framework based on alternating iterations of shrinkage thresholds, alternately updating blurring kernels and images, with the theoretical foundation of network design. Additionally, we propose a learnable blur kernel proximal mapping module to improve the blur kernel evaluation in the kernel domain. Then, we proposed a deep proximal mapping module in the image domain, which combines a generalized shrinkage threshold operator and a multi-scale prior feature extraction block. This module also introduces an attention mechanism to adaptively adjust the prior importance, thus avoiding the drawbacks of hand-crafted image prior terms. Thus, a novel multi-scale generalized shrinkage threshold network (MGSTNet) is designed to specifically focus on learning deep geometric prior features to enhance image restoration. Experiments demonstrate the superiority of our MGSTNet framework on remote sensing image datasets compared to existing deblurring methods.

Abstract: An underlying mechanism for successful deep learning (DL) with a limited deep architecture and dataset, namely VGG-16 on CIFAR-10, was recently presented based on a quantitative method to measure the quality of a single filter in each layer. In this method, each filter identifies small clusters of possible output labels, with additional noise selected as labels out of the clusters. This feature is progressively sharpened with the layers, resulting in an enhanced signal-to-noise ratio (SNR) and higher accuracy. In this study, the suggested universal mechanism is verified for VGG-16 and EfficientNet-B0 trained on the CIFAR-100 and ImageNet datasets with the following main results. First, the accuracy progressively increases with the layers, whereas the noise per filter typically progressively decreases. Second, for a given deep architecture, the maximal error rate increases approximately linearly with the number of output labels. Third, the average filter cluster size and the number of clusters per filter at the last convolutional layer adjacent to the output layer are almost independent of the number of dataset labels in the range [3, 1,000], while a high SNR is preserved. The presented DL mechanism suggests several techniques, such as applying filter's cluster connections (AFCC), to improve the computational complexity and accuracy of deep architectures and furthermore pinpoints the simplification of pre-existing structures while maintaining their accuracies.

Abstract: Road disease detection is challenging due to the the small proportion of road damage in target region and the diverse background,which introduce lots of domain information.Besides, disease categories have high similarity,makes the detection more difficult. In this paper, we propose a new LDBFSS(Latent Domain Background Feature Separation and Suppression) network which could perform background information separation and suppression without domain supervision and contrastive enhancement of object features.We combine our LDBFSS network with YOLOv5 model to enhance disease features for better road disease detection. As the components of LDBFSS network, we first design a latent domain discovery module and a domain adversarial learning module to obtain pseudo domain labels through unsupervised method, guiding domain discriminator and model to train adversarially to suppress background information. In addition, we introduce a contrastive learning module and design k-instance contrastive loss, optimize the disease feature representation by increasing the inter-class distance and reducing the intra-class distance for object features. We conducted experiments on two road disease detection datasets, GRDDC and CNRDD, and compared with other models,which show an increase of nearly 4% on GRDDC dataset compared with optimal model, and an increase of 4.6% on CNRDD dataset. Experimental results prove the effectiveness and superiority of our model.

Abstract: Large Language Models (LLMs), primarily trained on text-based datasets, exhibit exceptional proficiencies in understanding and executing complex linguistic instructions via text outputs. However, they falter when requests to generate non-text ones. Concurrently, modality conversion models, such as text-to-image, despite generating high-quality images, suffer from a lack of extensive textual pretraining. As a result, these models are only capable of accommodating specific image descriptions rather than comprehending more complex instructions. To bridge this gap, we propose a novel approach, \methodname, from a modality conversion perspective that evolves a text-based LLM into a multi-modal one. We specifically employ a minimal dataset to instruct LLMs to recognize the intended output modality as directed by the instructions. Consequently, the adapted LLM can effectively summon various off-the-shelf modality conversion models from the model zoos to generate non-text responses. This circumvents the necessity for complicated pretraining that typically requires immense quantities of paired multi-modal data, while simultaneously inheriting the extensive knowledge of LLMs and the ability of high-quality generative models. To evaluate and compare the adapted multi-modal LLM with its traditional counterparts, we have constructed a multi-modal instruction benchmark that solicits diverse modality outputs. The experiment results reveal that, with minimal training, LLMs can be conveniently adapted to comprehend requests for non-text responses, thus achieving higher flexibility in multi-modal scenarios. Code and data will be made available at https://github.com/xinke-wang/SwitchGPT.

Abstract: The reconstruction of indoor scenes from multi-view RGB images is challenging due to the coexistence of flat and texture-less regions alongside delicate and fine-grained regions. Recent methods leverage neural radiance fields aided by predicted surface normal priors to recover the scene geometry. These methods excel in producing complete and smooth results for floor and wall areas. However, they struggle to capture complex surfaces with high-frequency structures due to the inadequate neural representation and the inaccurately predicted normal priors. To improve the capacity of the implicit representation, we propose a hybrid architecture to represent low-frequency and high-frequency regions separately. To enhance the normal priors, we introduce a simple yet effective image sharpening and denoising technique, coupled with a network that estimates the pixel-wise uncertainty of the predicted surface normal vectors. Identifying such uncertainty can prevent our model from being misled by unreliable surface normal supervisions that hinder the accurate reconstruction of intricate geometries. Experiments on the benchmark datasets show that our method significantly outperforms existing methods in terms of reconstruction quality.

Abstract: Estimating the head pose of a person is a crucial problem for numerous applications that is yet mainly addressed as a subtask of frontal pose prediction. We present a novel method for unconstrained end-to-end head pose estimation to tackle the challenging task of full range of orientation head pose prediction. We address the issue of ambiguous rotation labels by introducing the rotation matrix formalism for our ground truth data and propose a continuous 6D rotation matrix representation for efficient and robust direct regression. This allows to efficiently learn full rotation appearance and to overcome the limitations of the current state-of-the-art. Together with new accumulated training data that provides full head pose rotation data and a geodesic loss approach for stable learning, we design an advanced model that is able to predict an extended range of head orientations. An extensive evaluation on public datasets demonstrates that our method significantly outperforms other state-of-the-art methods in an efficient and robust manner, while its advanced prediction range allows the expansion of the application area. We open-source our training and testing code along with our trained models: https://github.com/thohemp/6DRepNet360.

Abstract: Neural radiance field (NeRF) based methods enable high-quality novel-view synthesis for multi-view images. This work presents a method for synthesizing colorized novel views from input grey-scale multi-view images. When we apply image or video-based colorization methods on the generated grey-scale novel views, we observe artifacts due to inconsistency across views. Training a radiance field network on the colorized grey-scale image sequence also does not solve the 3D consistency issue. We propose a distillation based method to transfer color knowledge from the colorization networks trained on natural images to the radiance field network. Specifically, our method uses the radiance field network as a 3D representation and transfers knowledge from existing 2D colorization methods. The experimental results demonstrate that the proposed method produces superior colorized novel views for indoor and outdoor scenes while maintaining cross-view consistency than baselines. Further, we show the efficacy of our method on applications like colorization of radiance field network trained from 1.) Infra-Red (IR) multi-view images and 2.) Old grey-scale multi-view image sequences.

Abstract: Dataset condensation aims to condense a large dataset with a lot of training samples into a small set. Previous methods usually condense the dataset into the pixels format. However, it suffers from slow optimization speed and large number of parameters to be optimized. When increasing image resolutions and classes, the number of learnable parameters grows accordingly, prohibiting condensation methods from scaling up to large datasets with diverse classes. Moreover, the relations among condensed samples have been neglected and hence the feature distribution of condensed samples is often not diverse. To solve these problems, we propose to condense the dataset into another format, a generative model. Such a novel format allows for the condensation of large datasets because the size of the generative model remains relatively stable as the number of classes or image resolution increases. Furthermore, an intra-class and an inter-class loss are proposed to model the relation of condensed samples. Intra-class loss aims to create more diverse samples for each class by pushing each sample away from the others of the same class. Meanwhile, inter-class loss increases the discriminability of samples by widening the gap between the centers of different classes. Extensive comparisons with state-of-the-art methods and our ablation studies confirm the effectiveness of our method and its individual component. To our best knowledge, we are the first to successfully conduct condensation on ImageNet-1k.

Abstract: Accurate dietary intake estimation is critical for informing policies and programs to support healthy eating, as malnutrition has been directly linked to decreased quality of life. However self-reporting methods such as food diaries suffer from substantial bias. Other conventional dietary assessment techniques and emerging alternative approaches such as mobile applications incur high time costs and may necessitate trained personnel. Recent work has focused on using computer vision and machine learning to automatically estimate dietary intake from food images, but the lack of comprehensive datasets with diverse viewpoints, modalities and food annotations hinders the accuracy and realism of such methods. To address this limitation, we introduce NutritionVerse-Synth, the first large-scale dataset of 84,984 photorealistic synthetic 2D food images with associated dietary information and multimodal annotations (including depth images, instance masks, and semantic masks). Additionally, we collect a real image dataset, NutritionVerse-Real, containing 889 images of 251 dishes to evaluate realism. Leveraging these novel datasets, we develop and benchmark NutritionVerse, an empirical study of various dietary intake estimation approaches, including indirect segmentation-based and direct prediction networks. We further fine-tune models pretrained on synthetic data with real images to provide insights into the fusion of synthetic and real data. Finally, we release both datasets (NutritionVerse-Synth, NutritionVerse-Real) on https://www.kaggle.com/nutritionverse/datasets as part of an open initiative to accelerate machine learning for dietary sensing.

Abstract: Video matting has broad applications, from adding interesting effects to casually captured movies to assisting video production professionals. Matting with associated effects such as shadows and reflections has also attracted increasing research activity, and methods like Omnimatte have been proposed to separate dynamic foreground objects of interest into their own layers. However, prior works represent video backgrounds as 2D image layers, limiting their capacity to express more complicated scenes, thus hindering application to real-world videos. In this paper, we propose a novel video matting method, OmnimatteRF, that combines dynamic 2D foreground layers and a 3D background model. The 2D layers preserve the details of the subjects, while the 3D background robustly reconstructs scenes in real-world videos. Extensive experiments demonstrate that our method reconstructs scenes with better quality on various videos.

Abstract: In this paper, we present the decomposed triplane-hash neural radiance fields (DT-NeRF), a framework that significantly improves the photorealistic rendering of talking faces and achieves state-of-the-art results on key evaluation datasets. Our architecture decomposes the facial region into two specialized triplanes: one specialized for representing the mouth, and the other for the broader facial features. We introduce audio features as residual terms and integrate them as query vectors into our model through an audio-mouth-face transformer. Additionally, our method leverages the capabilities of Neural Radiance Fields (NeRF) to enrich the volumetric representation of the entire face through additive volumetric rendering techniques. Comprehensive experimental evaluations corroborate the effectiveness and superiority of our proposed approach.

Abstract: Given a group of images, co-salient object detection (CoSOD) aims to highlight the common salient object in each image. There are two factors closely related to the success of this task, namely consensus extraction, and the dispersion of consensus to each image. Most previous works represent the group consensus using local features, while we instead utilize a hierarchical Transformer module for extracting semantic-level consensus. Therefore, it can obtain a more comprehensive representation of the common object category, and exclude interference from other objects that share local similarities with the target object. In addition, we propose a Transformer-based dispersion module that takes into account the variation of the co-salient object in different scenes. It distributes the consensus to the image feature maps in an image-specific way while making full use of interactions within the group. These two modules are integrated with a ViT encoder and an FPN-like decoder to form an end-to-end trainable network, without additional branch and auxiliary loss. The proposed method is evaluated on three commonly used CoSOD datasets and achieves state-of-the-art performance.

Abstract: Large pre-trained vision-language models such as CLIP have demonstrated great potential in zero-shot transferability to downstream tasks. However, to attain optimal performance, the manual selection of prompts is necessary to improve alignment between the downstream image distribution and the textual class descriptions. This manual prompt engineering is the major challenge for deploying such models in practice since it requires domain expertise and is extremely time-consuming. To avoid non-trivial prompt engineering, recent work Context Optimization (CoOp) introduced the concept of prompt learning to the vision domain using learnable textual tokens. While CoOp can achieve substantial improvements over manual prompts, its learned context is worse generalizable to wider unseen classes within the same dataset. In this work, we present Prompt Learning with Reparameterization Encoder (PRE) - a simple and efficient method that enhances the generalization ability of the learnable prompt to unseen classes while maintaining the capacity to learn Base classes. Instead of directly optimizing the prompts, PRE employs a prompt encoder to reparameterize the input prompt embeddings, enhancing the exploration of task-specific knowledge from few-shot samples. Experiments and extensive ablation studies on 8 benchmarks demonstrate that our approach is an efficient method for prompt learning. Specifically, PRE achieves a notable enhancement of 5.60% in average accuracy on New classes and 3% in Harmonic mean compared to CoOp in the 16-shot setting, all achieved within a good training time.

Abstract: Previous evaluations on 6DoF object pose tracking have presented obvious limitations along with the development of this area. In particular, the evaluation protocols are not unified for different methods, the widely-used YCBV dataset contains significant annotation error, and the error metrics also may be biased. As a result, it is hard to fairly compare the methods, which has became a big obstacle for developing new algorithms. In this paper we contribute a unified benchmark to address the above problems. For more accurate annotation of YCBV, we propose a multi-view multi-object global pose refinement method, which can jointly refine the poses of all objects and view cameras, resulting in sub-pixel sub-millimeter alignment errors. The limitations of previous scoring methods and error metrics are analyzed, based on which we introduce our improved evaluation methods. The unified benchmark takes both YCBV and BCOT as base datasets, which are shown to be complementary in scene categories. In experiments, we validate the precision and reliability of the proposed global pose refinement method with a realistic semi-synthesized dataset particularly for YCBV, and then present the benchmark results unifying learning&non-learning and RGB&RGBD methods, with some finds not discovered in previous studies.

Abstract: More research attention has recently been given to end-to-end autonomous driving technologies where the entire driving pipeline is replaced with a single neural network because of its simpler structure and faster inference time. Despite this appealing approach largely reducing the components in driving pipeline, its simplicity also leads to interpretability problems and safety issues arXiv:2003.06404. The trained policy is not always compliant with the traffic rules and it is also hard to discover the reason for the misbehavior because of the lack of intermediate outputs. Meanwhile, Sensors are also critical to autonomous driving's security and feasibility to perceive the surrounding environment under complex driving scenarios. In this paper, we proposed P-CSG, a novel penalty-based imitation learning approach with cross semantics generation sensor fusion technologies to increase the overall performance of End-to-End Autonomous Driving. We conducted an assessment of our model's performance using the Town 05 Long benchmark, achieving an impressive driving score improvement of over 15%. Furthermore, we conducted robustness evaluations against adversarial attacks like FGSM and Dot attacks, revealing a substantial increase in robustness compared to baseline models.More detailed information, such as code-based resources, ablation studies and videos can be found at https://hk-zh.github.io/p-csg-plus.

Abstract: One of the main issues in computing a tensor decomposition is how to choose the number of rank-one components, since there is no finite algorithms for determining the rank of a tensor. A commonly used approach for this purpose is to find a low-dimensional subspace by solving an optimization problem and assuming the number of components is fixed. However, even though this algorithm is efficient and easy to implement, it often converges to poor local minima and suffers from outliers and noise. The aim of this paper is to develop a mathematical framework for exact tensor decomposition that is able to represent a tensor as the sum of a finite number of low-rank tensors. In the paper three different problems will be carried out to derive: i) the decomposition of a non-negative self-adjoint tensor operator; ii) the decomposition of a linear tensor transformation; iii) the decomposition of a generic tensor.

Abstract: Automatic road extraction from satellite imagery using deep learning is a viable alternative to traditional manual mapping. Therefore it has received considerable attention recently. However, most of the existing methods are supervised and require pixel-level labeling, which is tedious and error-prone. To make matters worse, the earth has a diverse range of terrain, vegetation, and man-made objects. It is well known that models trained in one area generalize poorly to other areas. Various shooting conditions such as light and angel, as well as different image processing techniques further complicate the issue. It is impractical to develop training data to cover all image styles. This paper proposes to leverage OpenStreetMap road data as weak labels and large scale satellite imagery to pre-train semantic segmentation models. Our extensive experimental results show that the prediction accuracy increases with the amount of the weakly labeled data, as well as the road density in the areas chosen for training. Using as much as 100 times more data than the widely used DeepGlobe road dataset, our model with the D-LinkNet architecture and the ResNet-50 backbone exceeds the top performer of the current DeepGlobe leaderboard. Furthermore, due to large-scale pre-training, our model generalizes much better than those trained with only the curated datasets, implying great application potential.

Abstract: Neural Radiance Fields (NeRF) employ multi-view images for 3D scene representation and have shown remarkable performance. As one of the primary sources of multi-view images, multi-camera systems encounter challenges such as varying intrinsic parameters and frequent pose changes. Most previous NeRF-based methods often assume a global unique camera and seldom consider scenarios with multiple cameras. Besides, some pose-robust methods still remain susceptible to suboptimal solutions when poses are poor initialized. In this paper, we propose MC-NeRF, a method can jointly optimize both intrinsic and extrinsic parameters for bundle-adjusting Neural Radiance Fields. Firstly, we conduct a theoretical analysis to tackle the degenerate case and coupling issue that arise from the joint optimization between intrinsic and extrinsic parameters. Secondly, based on the proposed solutions, we introduce an efficient calibration image acquisition scheme for multi-camera systems, including the design of calibration object. Lastly, we present a global end-to-end network with training sequence that enables the regression of intrinsic and extrinsic parameters, along with the rendering network. Moreover, most existing datasets are designed for unique camera, we create a new dataset that includes four different styles of multi-camera acquisition systems, allowing readers to generate custom datasets. Experiments confirm the effectiveness of our method when each image corresponds to different camera parameters. Specifically, we adopt up to 110 images with 110 different intrinsic and extrinsic parameters, to achieve 3D scene representation without providing initial poses. The Code and supplementary materials are available at https://in2-viaun.github.io/MC-NeRF.

Abstract: LiDAR semantic segmentation plays a crucial role in enabling autonomous driving and robots to understand their surroundings accurately and robustly. There are different types of methods, such as point-based, range image-based, and polar-based. Among these, range image-based methods are widely used due to their balance between accuracy and speed. However, they face a significant challenge known as the ``many-to-one'' problem caused by the range image's limited horizontal and vertical angular resolution, where around 20% of the 3D points are occluded during model inference based on our observation. In this paper, we present TFNet, a range image-based LiDAR semantic segmentation method that utilizes temporal information to address this issue. Specifically, we incorporate a temporal fusion layer to extract useful information from previous scans and integrate it with the current scan. We then design a max-voting-based post-processing technique to correct false predictions, particularly those caused by the ``many-to-one'' issue. Experiments on two benchmarks and seven backbones of three modalities demonstrate the effectiveness and scalability of our proposed method.

Abstract: This paper targets a novel trade-off problem in generalizable prompt learning for vision-language models (VLM), i.e., improving the performance on unseen classes while maintaining the performance on seen classes. Comparing with existing generalizable methods that neglect the seen classes degradation, the setting of this problem is more strict and fits more closely with practical applications. To solve this problem, we start from the optimization perspective, and leverage the relationship between loss landscape geometry and model generalization ability. By analyzing the loss landscape of the state-of-the-art method and the widely-used Sharpness-aware Minimization (SAM), we conclude that the trade-off performance correlates to both loss value and loss sharpness, while each of them are indispensable. However, we find the optimizing gradient of existing methods cannot always maintain high consistency with both loss value and loss sharpness during the whole optimization procedure. To this end, we propose an novel SAM-based method for prompt learning, denoted as Gradient Constrained Sharpness-aware Context Optimization (GCSCoOp), to dynamically constrains the optimizing gradient, thus achieving above two-fold optimization objective simultaneously. Extensive experiments verify the effectiveness of GCSCoOp in the trade-off problem.

Abstract: This work introduces a new multispectral database and novel approaches for eyeblink detection in RGB and Near-Infrared (NIR) individual images. Our contributed dataset (mEBAL2, multimodal Eye Blink and Attention Level estimation, Version 2) is the largest existing eyeblink database, representing a great opportunity to improve data-driven multispectral approaches for blink detection and related applications (e.g., attention level estimation and presentation attack detection in face biometrics). mEBAL2 includes 21,100 image sequences from 180 different students (more than 2 million labeled images in total) while conducting a number of e-learning tasks of varying difficulty or taking a real course on HTML initiation through the edX MOOC platform. mEBAL2 uses multiple sensors, including two Near-Infrared (NIR) and one RGB camera to capture facial gestures during the execution of the tasks, as well as an Electroencephalogram (EEG) band to get the cognitive activity of the user and blinking events. Furthermore, this work proposes a Convolutional Neural Network architecture as benchmark for blink detection on mEBAL2 with performances up to 97%. Different training methodologies are implemented using the RGB spectrum, NIR spectrum, and the combination of both to enhance the performance on existing eyeblink detectors. We demonstrate that combining NIR and RGB images during training improves the performance of RGB eyeblink detectors (i.e., detection based only on a RGB image). Finally, the generalization capacity of the proposed eyeblink detectors is validated in wilder and more challenging environments like the HUST-LEBW dataset to show the usefulness of mEBAL2 to train a new generation of data-driven approaches for eyeblink detection.

Abstract: We present a novel local-global feature fusion framework for body-weight exercise recognition with floor-based dynamic pressure maps. One step further from the existing studies using deep neural networks mainly focusing on global feature extraction, the proposed framework aims to combine local and global features using image processing techniques and the YOLO object detection to localize pressure profiles from different body parts and consider physical constraints. The proposed local feature extraction method generates two sets of high-level local features consisting of cropped pressure mapping and numerical features such as angular orientation, location on the mat, and pressure area. In addition, we adopt a knowledge distillation for regularization to preserve the knowledge of the global feature extraction and improve the performance of the exercise recognition. Our experimental results demonstrate a notable 11 percent improvement in F1 score for exercise recognition while preserving label-specific features.

Abstract: This paper presents a method to learn hand-object interaction prior for reconstructing a 3D hand-object scene from a single RGB image. The inference as well as training-data generation for 3D hand-object scene reconstruction is challenging due to the depth ambiguity of a single image and occlusions by the hand and object. We turn this challenge into an opportunity by utilizing the hand shape to constrain the possible relative configuration of the hand and object geometry. We design a generalizable implicit function, HandNeRF, that explicitly encodes the correlation of the 3D hand shape features and 2D object features to predict the hand and object scene geometry. With experiments on real-world datasets, we show that HandNeRF is able to reconstruct hand-object scenes of novel grasp configurations more accurately than comparable methods. Moreover, we demonstrate that object reconstruction from HandNeRF ensures more accurate execution of a downstream task, such as grasping for robotic hand-over.

Abstract: We present an approach to modeling an image-space prior on scene dynamics. Our prior is learned from a collection of motion trajectories extracted from real video sequences containing natural, oscillating motion such as trees, flowers, candles, and clothes blowing in the wind. Given a single image, our trained model uses a frequency-coordinated diffusion sampling process to predict a per-pixel long-term motion representation in the Fourier domain, which we call a neural stochastic motion texture. This representation can be converted into dense motion trajectories that span an entire video. Along with an image-based rendering module, these trajectories can be used for a number of downstream applications, such as turning still images into seamlessly looping dynamic videos, or allowing users to realistically interact with objects in real pictures.

Abstract: Existing volumetric methods for predicting 3D human pose estimation are accurate, but computationally expensive and optimized for single time-step prediction. We present TEMPO, an efficient multi-view pose estimation model that learns a robust spatiotemporal representation, improving pose accuracy while also tracking and forecasting human pose. We significantly reduce computation compared to the state-of-the-art by recurrently computing per-person 2D pose features, fusing both spatial and temporal information into a single representation. In doing so, our model is able to use spatiotemporal context to predict more accurate human poses without sacrificing efficiency. We further use this representation to track human poses over time as well as predict future poses. Finally, we demonstrate that our model is able to generalize across datasets without scene-specific fine-tuning. TEMPO achieves 10$\%$ better MPJPE with a 33$\times$ improvement in FPS compared to TesseTrack on the challenging CMU Panoptic Studio dataset.

Abstract: Recently, large-scale pre-trained language-image models like CLIP have shown extraordinary capabilities for understanding spatial contents, but naively transferring such models to video recognition still suffers from unsatisfactory temporal modeling capabilities. Existing methods insert tunable structures into or in parallel with the pre-trained model, which either requires back-propagation through the whole pre-trained model and is thus resource-demanding, or is limited by the temporal reasoning capability of the pre-trained structure. In this work, we present DiST, which disentangles the learning of spatial and temporal aspects of videos. Specifically, DiST uses a dual-encoder structure, where a pre-trained foundation model acts as the spatial encoder, and a lightweight network is introduced as the temporal encoder. An integration branch is inserted between the encoders to fuse spatio-temporal information. The disentangled spatial and temporal learning in DiST is highly efficient because it avoids the back-propagation of massive pre-trained parameters. Meanwhile, we empirically show that disentangled learning with an extra network for integration benefits both spatial and temporal understanding. Extensive experiments on five benchmarks show that DiST delivers better performance than existing state-of-the-art methods by convincing gaps. When pre-training on the large-scale Kinetics-710, we achieve 89.7% on Kinetics-400 with a frozen ViT-L model, which verifies the scalability of DiST. Codes and models can be found in https://github.com/alibaba-mmai-research/DiST.

Abstract: Object detection (OD), a crucial vision task, remains challenged by the lack of large training datasets with precise object localization labels. In this work, we propose ALWOD, a new framework that addresses this problem by fusing active learning (AL) with weakly and semi-supervised object detection paradigms. Because the performance of AL critically depends on the model initialization, we propose a new auxiliary image generator strategy that utilizes an extremely small labeled set, coupled with a large weakly tagged set of images, as a warm-start for AL. We then propose a new AL acquisition function, another critical factor in AL success, that leverages the student-teacher OD pair disagreement and uncertainty to effectively propose the most informative images to annotate. Finally, to complete the AL loop, we introduce a new labeling task delegated to human annotators, based on selection and correction of model-proposed detections, which is both rapid and effective in labeling the informative images. We demonstrate, across several challenging benchmarks, that ALWOD significantly narrows the gap between the ODs trained on few partially labeled but strategically selected image instances and those that rely on the fully-labeled data. Our code is publicly available on https://github.com/seqam-lab/ALWOD.

Abstract: While text-conditional 3D object generation and manipulation have seen rapid progress, the evaluation of coherence between generated 3D shapes and input textual descriptions lacks a clear benchmark. The reason is twofold: a) the low quality of the textual descriptions in the only publicly available dataset of text-shape pairs; b) the limited effectiveness of the metrics used to quantitatively assess such coherence. In this paper, we propose a comprehensive solution that addresses both weaknesses. Firstly, we employ large language models to automatically refine textual descriptions associated with shapes. Secondly, we propose a quantitative metric to assess text-to-shape coherence, through cross-attention mechanisms. To validate our approach, we conduct a user study and compare quantitatively our metric with existing ones. The refined dataset, the new metric and a set of text-shape pairs validated by the user study comprise a novel, fine-grained benchmark that we publicly release to foster research on text-to-shape coherence of text-conditioned 3D generative models. Benchmark available at https://cvlab-unibo.github.io/CrossCoherence-Web/.

Abstract: Human-Scene Interaction (HSI) is a vital component of fields like embodied AI and virtual reality. Despite advancements in motion quality and physical plausibility, two pivotal factors, versatile interaction control and the development of a user-friendly interface, require further exploration before the practical application of HSI. This paper presents a unified HSI framework, UniHSI, which supports unified control of diverse interactions through language commands. This framework is built upon the definition of interaction as Chain of Contacts (CoC): steps of human joint-object part pairs, which is inspired by the strong correlation between interaction types and human-object contact regions. Based on the definition, UniHSI constitutes a Large Language Model (LLM) Planner to translate language prompts into task plans in the form of CoC, and a Unified Controller that turns CoC into uniform task execution. To facilitate training and evaluation, we collect a new dataset named ScenePlan that encompasses thousands of task plans generated by LLMs based on diverse scenarios. Comprehensive experiments demonstrate the effectiveness of our framework in versatile task execution and generalizability to real scanned scenes. The project page is at https://github.com/OpenRobotLab/UniHSI .

Abstract: We introduce OpenIllumination, a real-world dataset containing over 108K images of 64 objects with diverse materials, captured under 72 camera views and a large number of different illuminations. For each image in the dataset, we provide accurate camera parameters, illumination ground truth, and foreground segmentation masks. Our dataset enables the quantitative evaluation of most inverse rendering and material decomposition methods for real objects. We examine several state-of-the-art inverse rendering methods on our dataset and compare their performances. The dataset and code can be found on the project page: https://oppo-us-research.github.io/OpenIllumination.

Abstract: Creating diverse and high-quality 3D assets with an automatic generative model is highly desirable. Despite extensive efforts on 3D generation, most existing works focus on the generation of a single category or a few categories. In this paper, we introduce a diffusion-based feed-forward framework for synthesizing massive categories of real-world 3D objects with a single generative model. Notably, there are three major challenges for this large-vocabulary 3D generation: a) the need for expressive yet efficient 3D representation; b) large diversity in geometry and texture across categories; c) complexity in the appearances of real-world objects. To this end, we propose a novel triplane-based 3D-aware Diffusion model with TransFormer, DiffTF, for handling challenges via three aspects. 1) Considering efficiency and robustness, we adopt a revised triplane representation and improve the fitting speed and accuracy. 2) To handle the drastic variations in geometry and texture, we regard the features of all 3D objects as a combination of generalized 3D knowledge and specialized 3D features. To extract generalized 3D knowledge from diverse categories, we propose a novel 3D-aware transformer with shared cross-plane attention. It learns the cross-plane relations across different planes and aggregates the generalized 3D knowledge with specialized 3D features. 3) In addition, we devise the 3D-aware encoder/decoder to enhance the generalized 3D knowledge in the encoded triplanes for handling categories with complex appearances. Extensive experiments on ShapeNet and OmniObject3D (over 200 diverse real-world categories) convincingly demonstrate that a single DiffTF model achieves state-of-the-art large-vocabulary 3D object generation performance with large diversity, rich semantics, and high quality.