Computer Vision and Pattern Recognition (cs.CV)

Abstract: Restricted Boltzmann Machines are generative models that consist of a layer of hidden variables connected to another layer of visible units, and they are used to model the distribution over visible variables. In order to gain a higher representability power, many hidden units are commonly used, which, in combination with a large number of visible units, leads to a high number of trainable parameters. In this work we introduce the Structural Restricted Boltzmann Machine model, which taking advantage of the structure of the data in hand, constrains connections of hidden units to subsets of visible units in order to reduce significantly the number of trainable parameters, without compromising performance. As a possible area of application, we focus on image modelling. Based on the nature of the images, the structure of the connections is given in terms of spatial neighbourhoods over the pixels of the image that constitute the visible variables of the model. We conduct extensive experiments on various image domains. Image denoising is evaluated with corrupted images from the MNIST dataset. The generative power of our models is compared to vanilla RBMs, as well as their classification performance, which is assessed with five different image domains. Results show that our proposed model has a faster and more stable training, while also obtaining better results compared to an RBM with no constrained connections between its visible and hidden units.

Abstract: The color video inpainting problem is one of the most challenging problem in the modern imaging science. It aims to recover a color video from a small part of pixels that may contain noise. However, there are less of robust models that can simultaneously preserve the coupling of color channels and the evolution of color video frames. In this paper, we present a new robust quaternion tensor completion (RQTC) model to solve this challenging problem and derive the exact recovery theory. The main idea is to build a quaternion tensor optimization model to recover a low-rank quaternion tensor that represents the targeted color video and a sparse quaternion tensor that represents noise. This new model is very efficient to recover high dimensional data that satisfies the prior low-rank assumption. To solve the case without low-rank property, we introduce a new low-rank learning RQTC model, which rearranges similar patches classified by a quaternion learning method into smaller tensors satisfying the prior low-rank assumption. We also propose fast algorithms with global convergence guarantees. In numerical experiments, the proposed methods successfully recover color videos with eliminating color contamination and keeping the continuity of video scenery, and their solutions are of higher quality in terms of PSNR and SSIM values than the state-of-the-art algorithms.

Abstract: Humans possess the cognitive ability to comprehend scenes in a compositional manner. To empower AI systems with similar abilities, object-centric representation learning aims to acquire representations of individual objects from visual scenes without any supervision. Although recent advancements in object-centric representation learning have achieved remarkable progress on complex synthesis datasets, there is a huge challenge for application in complex real-world scenes. One of the essential reasons is the scarcity of real-world datasets specifically tailored to object-centric representation learning methods. To solve this problem, we propose a versatile real-world dataset of tabletop scenes for object-centric learning called OCTScenes, which is meticulously designed to serve as a benchmark for comparing, evaluating and analyzing object-centric representation learning methods. OCTScenes contains 5000 tabletop scenes with a total of 15 everyday objects. Each scene is captured in 60 frames covering a 360-degree perspective. Consequently, OCTScenes is a versatile benchmark dataset that can simultaneously satisfy the evaluation of object-centric representation learning methods across static scenes, dynamic scenes, and multi-view scenes tasks. Extensive experiments of object-centric representation learning methods for static, dynamic and multi-view scenes are conducted on OCTScenes. The results demonstrate the shortcomings of state-of-the-art methods for learning meaningful representations from real-world data, despite their impressive performance on complex synthesis datasets. Furthermore, OCTScenes can serves as a catalyst for advancing existing state-of-the-art methods, inspiring them to adapt to real-world scenes. Dataset and code are available at https://huggingface.co/datasets/Yinxuan/OCTScenes.

Abstract: Open-vocabulary object detection has benefited greatly from pretrained vision-language models, but is still limited by the amount of available detection training data. While detection training data can be expanded by using Web image-text pairs as weak supervision, this has not been done at scales comparable to image-level pretraining. Here, we scale up detection data with self-training, which uses an existing detector to generate pseudo-box annotations on image-text pairs. Major challenges in scaling self-training are the choice of label space, pseudo-annotation filtering, and training efficiency. We present the OWLv2 model and OWL-ST self-training recipe, which address these challenges. OWLv2 surpasses the performance of previous state-of-the-art open-vocabulary detectors already at comparable training scales (~10M examples). However, with OWL-ST, we can scale to over 1B examples, yielding further large improvement: With an L/14 architecture, OWL-ST improves AP on LVIS rare classes, for which the model has seen no human box annotations, from 31.2% to 44.6% (43% relative improvement). OWL-ST unlocks Web-scale training for open-world localization, similar to what has been seen for image classification and language modelling.

Abstract: Convolutional neural networks (CNNs) have recently proven their excellent ability to segment 2D cardiac ultrasound images. However, the majority of attempts to perform full-sequence segmentation of cardiac ultrasound videos either rely on models trained only on keyframe images or fail to maintain the topology over time. To address these issues, in this work, we consider segmentation of ultrasound video as a registration estimation problem and present a novel method for diffeomorphic image registration using neural ordinary differential equations (Neural ODE). In particular, we consider the registration field vector field between frames as a continuous trajectory ODE. The estimated registration field is then applied to the segmentation mask of the first frame to obtain a segment for the whole cardiac cycle. The proposed method, Echo-ODE, introduces several key improvements compared to the previous state-of-the-art. Firstly, by solving a continuous ODE, the proposed method achieves smoother segmentation, preserving the topology of segmentation maps over the whole sequence (Hausdorff distance: 3.7-4.4). Secondly, it maintains temporal consistency between frames without explicitly optimizing for temporal consistency attributes, achieving temporal consistency in 91% of the videos in the dataset. Lastly, the proposed method is able to maintain the clinical accuracy of the segmentation maps (MAE of the LVEF: 2.7-3.1). The results show that our method surpasses the previous state-of-the-art in multiple aspects, demonstrating the importance of spatial-temporal data processing for the implementation of Neural ODEs in medical imaging applications. These findings open up new research directions for solving echocardiography segmentation tasks.

Abstract: Vectorized high-definition map (HD-map) construction, which focuses on the perception of centimeter-level environmental information, has attracted significant research interest in the autonomous driving community. Most existing approaches first obtain rasterized map with the segmentation-based pipeline and then conduct heavy post-processing for downstream-friendly vectorization. In this paper, by delving into parameterization-based methods, we pioneer a concise and elegant scheme that adopts unified piecewise Bezier curve. In order to vectorize changeful map elements end-to-end, we elaborate a simple yet effective architecture, named Piecewise Bezier HD-map Network (BeMapNet), which is formulated as a direct set prediction paradigm and postprocessing-free. Concretely, we first introduce a novel IPM-PE Align module to inject 3D geometry prior into BEV features through common position encoding in Transformer. Then a well-designed Piecewise Bezier Head is proposed to output the details of each map element, including the coordinate of control points and the segment number of curves. In addition, based on the progressively restoration of Bezier curve, we also present an efficient Point-Curve-Region Loss for supervising more robust and precise HD-map modeling. Extensive comparisons show that our method is remarkably superior to other existing SOTAs by 18.0 mAP at least.

Abstract: Deep neural networks (DNNs) have been widely applied in medical image classification and achieve remarkable classification performance. These achievements heavily depend on large-scale accurately annotated training data. However, label noise is inevitably introduced in the medical image annotation, as the labeling process heavily relies on the expertise and experience of annotators. Meanwhile, DNNs suffer from overfitting noisy labels, degrading the performance of models. Therefore, in this work, we innovatively devise noise-robust training approach to mitigate the adverse effects of noisy labels in medical image classification. Specifically, we incorporate contrastive learning and intra-group attention mixup strategies into the vanilla supervised learning. The contrastive learning for feature extractor helps to enhance visual representation of DNNs. The intra-group attention mixup module constructs groups and assigns self-attention weights for group-wise samples, and subsequently interpolates massive noisy-suppressed samples through weighted mixup operation. We conduct comparative experiments on both synthetic and real-world noisy medical datasets under various noise levels. Rigorous experiments validate that our noise-robust method with contrastive learning and attention mixup can effectively handle with label noise, and is superior to state-of-the-art methods. An ablation study also shows that both components contribute to boost model performance. The proposed method demonstrates its capability of curb label noise and has certain potential toward real-world clinic applications.

Abstract: Pre-training & fine-tuning is a prevalent paradigm in computer vision (CV). Recently, parameter-efficient transfer learning (PETL) methods have shown promising performance in transferring knowledge from pre-trained models with only a few trainable parameters. Despite their success, the existing PETL methods in CV can be computationally expensive and require large amounts of memory and time cost during training, which limits low-resource users from conducting research and applications on large models. In this work, we propose Parameter, Memory, and Time Efficient Visual Adapter ($\mathrm{E^3VA}$) tuning to address this issue. We provide a gradient backpropagation highway for low-rank adapters which removes large gradient computations for the frozen pre-trained parameters, resulting in substantial savings of training memory and training time. Furthermore, we optimise the $\mathrm{E^3VA}$ structure for dense predictions tasks to promote model performance. Extensive experiments on COCO, ADE20K, and Pascal VOC benchmarks show that $\mathrm{E^3VA}$ can save up to 62.2% training memory and 26.2% training time on average, while achieving comparable performance to full fine-tuning and better performance than most PETL methods. Note that we can even train the Swin-Large-based Cascade Mask RCNN on GTX 1080Ti GPUs with less than 1.5% trainable parameters.

Abstract: Despite the notable accomplishments of deep object detection models, a major challenge that persists is the requirement for extensive amounts of training data. The process of procuring such real-world data is a laborious undertaking, which has prompted researchers to explore new avenues of research, such as synthetic data generation techniques. This study presents a framework for the generation of synthetic datasets by fine-tuning pretrained stable diffusion models. The synthetic datasets are then manually annotated and employed for training various object detection models. These detectors are evaluated on a real-world test set of 331 images and compared against a baseline model that was trained on real-world images. The results of this study reveal that the object detection models trained on synthetic data perform similarly to the baseline model. In the context of apple detection in orchards, the average precision deviation with the baseline ranges from 0.09 to 0.12. This study illustrates the potential of synthetic data generation techniques as a viable alternative to the collection of extensive training data for the training of deep models.

Abstract: Disaster mapping is a critical task that often requires on-site experts and is time-consuming. To address this, a comprehensive framework is presented for fast and accurate recognition of disasters using machine learning, termed DisasterNets. It consists of two stages, space granulation and attribute granulation. The space granulation stage leverages supervised/semi-supervised learning, unsupervised change detection, and domain adaptation with/without source data techniques to handle different disaster mapping scenarios. Furthermore, the disaster database with the corresponding geographic information field properties is built by using the attribute granulation stage. The framework is applied to earthquake-triggered landslide mapping and large-scale flood mapping. The results demonstrate a competitive performance for high-precision, high-efficiency, and cross-scene recognition of disasters. To bridge the gap between disaster mapping and machine learning communities, we will provide an openly accessible tool based on DisasterNets. The framework and tool will be available at https://github.com/HydroPML/DisasterNets.

Abstract: Pedestrian Attribute Recognition (PAR) deals with the problem of identifying features in a pedestrian image. It has found interesting applications in person retrieval, suspect re-identification and soft biometrics. In the past few years, several Deep Neural Networks (DNNs) have been designed to solve the task; however, the developed DNNs predominantly suffer from over-parameterization and high computational complexity. These problems hinder them from being exploited in resource-constrained embedded devices with limited memory and computational capacity. By reducing a network's layers using effective compression techniques, such as tensor decomposition, neural network compression is an effective method to tackle these problems. We propose novel Lightweight Attribute Localizing Models (LWALM) for Pedestrian Attribute Recognition (PAR). LWALM is a compressed neural network obtained after effective layer-wise compression of the Attribute Localization Model (ALM) using the Canonical Polyadic Decomposition with Error Preserving Correction (CPD-EPC) algorithm.

Abstract: With the current ubiquity of deep learning methods to solve computer vision and remote sensing specific tasks, the need for labelled data is growing constantly. However, in many cases, the annotation process can be long and tedious depending on the expertise needed to perform reliable annotations. In order to alleviate this need for annotations, several self-supervised methods have recently been proposed in the literature. The core principle behind these methods is to learn an image encoder using solely unlabelled data samples. In earth observation, there are opportunities to exploit domain-specific remote sensing image data in order to improve these methods. Specifically, by leveraging the geographical position associated with each image, it is possible to cross reference a location captured from multiple sensors, leading to multiple views of the same locations. In this paper, we briefly review the core principles behind so-called joint-embeddings methods and investigate the usage of multiple remote sensing modalities in self-supervised pre-training. We evaluate the final performance of the resulting encoders on the task of methane source classification.

Abstract: The lack of explainability limits the adoption of deep learning models in clinical practice. While methods exist to improve the understanding of such models, these are mainly saliency-based and developed for classification, despite many important tasks in medical imaging being continuous regression problems. Therefore, in this work, we present ExPeRT: an explainable prototype-based model specifically designed for regression tasks. Our proposed model makes a sample prediction from the distances to a set of learned prototypes in latent space, using a weighted mean of prototype labels. The distances in latent space are regularized to be relative to label differences, and each of the prototypes can be visualized as a sample from the training set. The image-level distances are further constructed from patch-level distances, in which the patches of both images are structurally matched using optimal transport. We demonstrate our proposed model on the task of brain age prediction on two image datasets: adult MR and fetal ultrasound. Our approach achieved state-of-the-art prediction performance while providing insight in the model's reasoning process.

Abstract: For a very long time, unsupervised learning for anomaly detection has been at the heart of image processing research and a stepping stone for high performance industrial automation process. With the emergence of CNN, several methods have been proposed such as Autoencoders, GAN, deep feature extraction, etc. In this paper, we propose a new method based on the promising concept of knowledge distillation which consists of training a network (the student) on normal samples while considering the output of a larger pretrained network (the teacher). The main contributions of this paper are twofold: First, a reduced student architecture with optimal layer selection is proposed, then a new Student-Teacher architecture with network bias reduction combining two teachers is proposed in order to jointly enhance the performance of anomaly detection and its localization accuracy. The proposed texture anomaly detector has an outstanding capability to detect defects in any texture and a fast inference time compared to the SOTA methods.

Abstract: We introduce AvatarBooth, a novel method for generating high-quality 3D avatars using text prompts or specific images. Unlike previous approaches that can only synthesize avatars based on simple text descriptions, our method enables the creation of personalized avatars from casually captured face or body images, while still supporting text-based model generation and editing. Our key contribution is the precise avatar generation control by using dual fine-tuned diffusion models separately for the human face and body. This enables us to capture intricate details of facial appearance, clothing, and accessories, resulting in highly realistic avatar generations. Furthermore, we introduce pose-consistent constraint to the optimization process to enhance the multi-view consistency of synthesized head images from the diffusion model and thus eliminate interference from uncontrolled human poses. In addition, we present a multi-resolution rendering strategy that facilitates coarse-to-fine supervision of 3D avatar generation, thereby enhancing the performance of the proposed system. The resulting avatar model can be further edited using additional text descriptions and driven by motion sequences. Experiments show that AvatarBooth outperforms previous text-to-3D methods in terms of rendering and geometric quality from either text prompts or specific images. Please check our project website at https://zeng-yifei.github.io/avatarbooth_page/.

Abstract: Despite the remarkable performance of text-to-image diffusion models in image generation tasks, recent studies have raised the issue that generated images sometimes cannot capture the intended semantic contents of the text prompts, which phenomenon is often called semantic misalignment. To address this, here we present a novel energy-based model (EBM) framework. Specifically, we first formulate EBMs of latent image representations and text embeddings in each cross-attention layer of the denoising autoencoder. Then, we obtain the gradient of the log posterior of context vectors, which can be updated and transferred to the subsequent cross-attention layer, thereby implicitly minimizing a nested hierarchy of energy functions. Our latent EBMs further allow zero-shot compositional generation as a linear combination of cross-attention outputs from different contexts. Using extensive experiments, we demonstrate that the proposed method is highly effective in handling various image generation tasks, including multi-concept generation, text-guided image inpainting, and real and synthetic image editing.

Abstract: Cloud detection is a pivotal satellite image pre-processing step that can be performed both on the ground and on board a satellite to tag useful images. In the latter case, it can reduce the amount of data to downlink by pruning the cloudy areas, or to make a satellite more autonomous through data-driven acquisition re-scheduling. We approach this task with nnU-Nets, a self-reconfigurable framework able to perform meta-learning of a segmentation network over various datasets. Unfortunately, such models are commonly memory-inefficient due to their (very) large architectures. To benefit from them in on-board processing, we compress nnU-Nets with knowledge distillation into much smaller and compact U-Nets. Our experiments, performed over Sentinel-2 and Landsat-8 images revealed that nnU-Nets deliver state-of-the-art performance without any manual design. Our approach was ranked within the top 7% best solutions (across 847 teams) in the On Cloud N: Cloud Cover Detection Challenge, where we reached the Jaccard index of 0.882 over more than 10k unseen Sentinel-2 images (the winners obtained 0.897, the baseline U-Net with the ResNet-34 backbone: 0.817, and the classic Sentinel-2 image thresholding: 0.652). Finally, we showed that knowledge distillation enables to elaborate dramatically smaller (almost 280x) U-Nets when compared to nnU-Nets while still maintaining their segmentation capabilities.

Abstract: With the advent of mobile phone photography and point-and-shoot cameras, deep-burst imaging is widely used for a number of photographic effects such as depth of field, super-resolution, motion deblurring, and image denoising. In this work, we propose to solve the problem of deep-burst image denoising by including an optical flow-based correspondence estimation module which aligns all the input burst images with respect to a reference frame. In order to deal with varying noise levels the individual burst images are pre-filtered with different settings. Exploiting the established correspondences one network block predicts a pixel-wise spatially-varying filter kernel to smooth each image in the original and prefiltered bursts before fusing all images to generate the final denoised output. The resulting pipeline achieves state-of-the-art results by combining all available information provided by the burst.

Abstract: Orthogonality regularization has been developed to prevent deep CNNs from training instability and feature redundancy. Among existing proposals, kernel orthogonality regularization enforces orthogonality by minimizing the residual between the Gram matrix formed by convolutional filters and the orthogonality matrix. We propose a novel measure for achieving better orthogonality among filters, which disentangles diagonal and correlation information from the residual. The model equipped with the measure under the principle of imposing strict orthogonality between filters surpasses previous regularization methods in near-orthogonality. Moreover, we observe the benefits of improved strict filter orthogonality in relatively shallow models, but as model depth increases, the performance gains in models employing strict kernel orthogonality decrease sharply. Furthermore, based on the observation of the potential conflict between strict kernel orthogonality and growing model capacity, we propose a relaxation theory on kernel orthogonality regularization. The relaxed kernel orthogonality achieves enhanced performance on models with increased capacity, shedding light on the burden of strict kernel orthogonality on deep model performance. We conduct extensive experiments with our kernel orthogonality regularization toolkit on ResNet and WideResNet in CIFAR-10 and CIFAR-100. We observe state-of-the-art gains in model performance from the toolkit, which includes both strict orthogonality and relaxed orthogonality regularization, and obtain more robust models with expressive features. These experiments demonstrate the efficacy of our toolkit and subtly provide insights into the often overlooked challenges posed by strict orthogonality, addressing the burden of strict orthogonality on capacity-rich models.

Abstract: Smart-parking solutions use sensors, cameras, and data analysis to improve parking efficiency and reduce traffic congestion. Computer vision-based methods have been used extensively in recent years to tackle the problem of parking lot management, but most of the works assume that the parking spots are manually labeled, impacting the cost and feasibility of deployment. To fill this gap, this work presents an automatic parking space detection method, which receives a sequence of images of a parking lot and returns a list of coordinates identifying the detected parking spaces. The proposed method employs instance segmentation to identify cars and, using vehicle occurrence, generate a heat map of parking spaces. The results using twelve different subsets from the PKLot and CNRPark-EXT parking lot datasets show that the method achieved an AP25 score up to 95.60\% and AP50 score up to 79.90\%.

Abstract: The robustness of image segmentation has been an important research topic in the past few years as segmentation models have reached production-level accuracy. However, like classification models, segmentation models can be vulnerable to adversarial perturbations, which hinders their use in critical-decision systems like healthcare or autonomous driving. Recently, randomized smoothing has been proposed to certify segmentation predictions by adding Gaussian noise to the input to obtain theoretical guarantees. However, this method exhibits a trade-off between the amount of added noise and the level of certification achieved. In this paper, we address the problem of certifying segmentation prediction using a combination of randomized smoothing and diffusion models. Our experiments show that combining randomized smoothing and diffusion models significantly improves certified robustness, with results indicating a mean improvement of 21 points in accuracy compared to previous state-of-the-art methods on Pascal-Context and Cityscapes public datasets. Our method is independent of the selected segmentation model and does not need any additional specialized training procedure.

Abstract: In this paper, we focus on the important yet understudied problem of Continual Federated Learning (CFL), where a server communicates with a set of clients to incrementally learn new concepts over time without sharing or storing any data. The complexity of this problem is compounded by challenges from both the Continual and Federated Learning perspectives. Specifically, models trained in a CFL setup suffer from catastrophic forgetting which is exacerbated by data heterogeneity across clients. Existing attempts at this problem tend to impose large overheads on clients and communication channels or require access to stored data which renders them unsuitable for real-world use due to privacy. In this paper, we attempt to tackle forgetting and heterogeneity while minimizing overhead costs and without requiring access to any stored data. We achieve this by leveraging a prompting based approach (such that only prompts and classifier heads have to be communicated) and proposing a novel and lightweight generation and distillation scheme to consolidate client models at the server. We formulate this problem for image classification and establish strong baselines for comparison, conduct experiments on CIFAR-100 as well as challenging, large-scale datasets like ImageNet-R and DomainNet. Our approach outperforms both existing methods and our own baselines by as much as 7% while significantly reducing communication and client-level computation costs.

Abstract: Visual question answering (VQA) is a challenging task that requires the ability to comprehend and reason with visual information. While recent vision-language models have made strides, they continue to struggle with zero-shot VQA, particularly in handling complex compositional questions and adapting to new domains i.e. knowledge-based reasoning. This paper explores the use of various prompting strategies, focusing on the BLIP2 model, to enhance zero-shot VQA performance. We conduct a comprehensive investigation across several VQA datasets, examining the effectiveness of different question templates, the role of few-shot exemplars, the impact of chain-of-thought (CoT) reasoning, and the benefits of incorporating image captions as additional visual cues. Despite the varied outcomes, our findings demonstrate that carefully designed question templates and the integration of additional visual cues, like image captions, can contribute to improved VQA performance, especially when used in conjunction with few-shot examples. However, we also identify a limitation in the use of chain-of-thought rationalization, which negatively affects VQA accuracy. Our study thus provides critical insights into the potential of prompting for improving zero-shot VQA performance.

Abstract: Data augmentation is known to improve the generalization capabilities of neural networks, provided that the set of transformations is chosen with care, a selection often performed manually. Automatic data augmentation aims at automating this process. However, most recent approaches still rely on some prior information; they start from a small pool of manually-selected default transformations that are either used to pretrain the network or forced to be part of the policy learned by the automatic data augmentation algorithm. In this paper, we propose to directly learn the augmentation policy without leveraging such prior knowledge. The resulting bilevel optimization problem becomes more challenging due to the larger search space and the inherent instability of bilevel optimization algorithms. To mitigate these issues (i) we follow a successive cold-start strategy with a Kullback-Leibler regularization, and (ii) we parameterize magnitudes as continuous distributions. Our approach leads to competitive results on standard benchmarks despite a more challenging setting, and generalizes beyond natural images.

Abstract: As neural networks become deeper, the redundancy within their parameters increases. This phenomenon has led to several methods that attempt to reduce the correlation between convolutional filters. We propose a computationally efficient regularization technique that encourages orthonormality between groups of filters within the same layer. Our experiments show that when incorporated into recent adaptation methods for diffusion models and vision transformers (ViTs), this regularization improves performance on downstream tasks. We further show improved robustness when group orthogonality is enforced during adversarial training. Our code is available at https://github.com/YoavKurtz/GOR.

Abstract: There is an emerging effort to combine the two popular technical paths, i.e., the multi-view stereo (MVS) and neural implicit surface (NIS), in scene reconstruction from sparse views. In this paper, we introduce a novel integration scheme that combines the multi-view stereo with neural signed distance function representations, which potentially overcomes the limitations of both methods. MVS uses per-view depth estimation and cross-view fusion to generate accurate surface, while NIS relies on a common coordinate volume. Based on this, we propose to construct per-view cost frustum for finer geometry estimation, and then fuse cross-view frustums and estimate the implicit signed distance functions to tackle noise and hole issues. We further apply a cascade frustum fusion strategy to effectively captures global-local information and structural consistency. Finally, we apply cascade sampling and a pseudo-geometric loss to foster stronger integration between the two architectures. Extensive experiments demonstrate that our method reconstructs robust surfaces and outperforms existing state-of-the-art methods.

Abstract: This paper presents a novel approach for text/speech-driven animation of a photo-realistic head model based on blend-shape geometry, dynamic textures, and neural rendering. Training a VAE for geometry and texture yields a parametric model for accurate capturing and realistic synthesis of facial expressions from a latent feature vector. Our animation method is based on a conditional CNN that transforms text or speech into a sequence of animation parameters. In contrast to previous approaches, our animation model learns disentangling/synthesizing different acting-styles in an unsupervised manner, requiring only phonetic labels that describe the content of training sequences. For realistic real-time rendering, we train a U-Net that refines rasterization-based renderings by computing improved pixel colors and a foreground matte. We compare our framework qualitatively/quantitatively against recent methods for head modeling as well as facial animation and evaluate the perceived rendering/animation quality in a user-study, which indicates large improvements compared to state-of-the-art approaches

Abstract: The success of deep learning based face recognition systems has given rise to serious privacy concerns due to their ability to enable unauthorized tracking of users in the digital world. Existing methods for enhancing privacy fail to generate naturalistic images that can protect facial privacy without compromising user experience. We propose a novel two-step approach for facial privacy protection that relies on finding adversarial latent codes in the low-dimensional manifold of a pretrained generative model. The first step inverts the given face image into the latent space and finetunes the generative model to achieve an accurate reconstruction of the given image from its latent code. This step produces a good initialization, aiding the generation of high-quality faces that resemble the given identity. Subsequently, user-defined makeup text prompts and identity-preserving regularization are used to guide the search for adversarial codes in the latent space. Extensive experiments demonstrate that faces generated by our approach have stronger black-box transferability with an absolute gain of 12.06% over the state-of-the-art facial privacy protection approach under the face verification task. Finally, we demonstrate the effectiveness of the proposed approach for commercial face recognition systems. Our code is available at https://github.com/fahadshamshad/Clip2Protect.

Abstract: Text-guided image editing is widely needed in daily life, ranging from personal use to professional applications such as Photoshop. However, existing methods are either zero-shot or trained on an automatically synthesized dataset, which contains a high volume of noise. Thus, they still require lots of manual tuning to produce desirable outcomes in practice. To address this issue, we introduce MagicBrush (https://osu-nlp-group.github.io/MagicBrush/), the first large-scale, manually annotated dataset for instruction-guided real image editing that covers diverse scenarios: single-turn, multi-turn, mask-provided, and mask-free editing. MagicBrush comprises over 10K manually annotated triples (source image, instruction, target image), which supports trainining large-scale text-guided image editing models. We fine-tune InstructPix2Pix on MagicBrush and show that the new model can produce much better images according to human evaluation. We further conduct extensive experiments to evaluate current image editing baselines from multiple dimensions including quantitative, qualitative, and human evaluations. The results reveal the challenging nature of our dataset and the gap between current baselines and real-world editing needs.

Abstract: Comprehensive modeling of the surrounding 3D world is key to the success of autonomous driving. However, existing perception tasks like object detection, road structure segmentation, depth & elevation estimation, and open-set object localization each only focus on a small facet of the holistic 3D scene understanding task. This divide-and-conquer strategy simplifies the algorithm development procedure at the cost of losing an end-to-end unified solution to the problem. In this work, we address this limitation by studying camera-based 3D panoptic segmentation, aiming to achieve a unified occupancy representation for camera-only 3D scene understanding. To achieve this, we introduce a novel method called PanoOcc, which utilizes voxel queries to aggregate spatiotemporal information from multi-frame and multi-view images in a coarse-to-fine scheme, integrating feature learning and scene representation into a unified occupancy representation. We have conducted extensive ablation studies to verify the effectiveness and efficiency of the proposed method. Our approach achieves new state-of-the-art results for camera-based semantic segmentation and panoptic segmentation on the nuScenes dataset. Furthermore, our method can be easily extended to dense occupancy prediction and has shown promising performance on the Occ3D benchmark. The code will be released at https://github.com/Robertwyq/PanoOcc.

Abstract: We propose a novel knowledge distillation framework for effectively teaching a sensorimotor student agent to drive from the supervision of a privileged teacher agent. Current distillation for sensorimotor agents methods tend to result in suboptimal learned driving behavior by the student, which we hypothesize is due to inherent differences between the input, modeling capacity, and optimization processes of the two agents. We develop a novel distillation scheme that can address these limitations and close the gap between the sensorimotor agent and its privileged teacher. Our key insight is to design a student which learns to align their input features with the teacher's privileged Bird's Eye View (BEV) space. The student then can benefit from direct supervision by the teacher over the internal representation learning. To scaffold the difficult sensorimotor learning task, the student model is optimized via a student-paced coaching mechanism with various auxiliary supervision. We further propose a high-capacity imitation learned privileged agent that surpasses prior privileged agents in CARLA and ensures the student learns safe driving behavior. Our proposed sensorimotor agent results in a robust image-based behavior cloning agent in CARLA, improving over current models by over 20.6% in driving score without requiring LiDAR, historical observations, ensemble of models, on-policy data aggregation or reinforcement learning.