Computer Vision and Pattern Recognition (cs.CV)

Abstract: We abstract the features~(\textit{i.e.} learned representations) of multi-modal data into 1)~\emph{uni-modal features}, which can be learned from uni-modal training, and 2) \emph{paired features}, which can \emph{only} be learned from cross-modal interactions. Multi-modal models are expected to benefit from cross-modal interactions on the basis of ensuring uni-modal feature learning. However, recent supervised multi-modal late-fusion training approaches still suffer from insufficient learning of uni-modal features on each modality. \emph{We prove that this phenomenon does hurt the model's generalization ability}. To this end, we propose to choose a targeted late-fusion learning method for the given supervised multi-modal task from \textbf{U}ni-\textbf{M}odal \textbf{E}nsemble~(UME) and the proposed \textbf{U}ni-\textbf{M}odal \textbf{T}eacher~(UMT), according to the distribution of uni-modal and paired features. We demonstrate that, under a simple guiding strategy, we can achieve comparable results to other complex late-fusion or intermediate-fusion methods on various multi-modal datasets, including VGG-Sound, Kinetics-400, UCF101, and ModelNet40.

Abstract: Compositional Zero-shot Learning (CZSL) aims to recognize novel concepts composed of known knowledge without training samples. Standard CZSL either identifies visual primitives or enhances unseen composed entities, and as a result, entanglement between state and object primitives cannot be fully utilized. Admittedly, vision-language models (VLMs) could naturally cope with CZSL through tuning prompts, while uneven entanglement leads prompts to be dragged into local optimum. In this paper, we take a further step to introduce a novel Disentangled and Recurrent Prompt Tuning framework termed DRPT to better tap the potential of VLMs in CZSL. Specifically, the state and object primitives are deemed as learnable tokens of vocabulary embedded in prompts and tuned on seen compositions. Instead of jointly tuning state and object, we devise a disentangled and recurrent tuning strategy to suppress the traction force caused by entanglement and gradually optimize the token parameters, leading to a better prompt space. Notably, we develop a progressive fine-tuning procedure that allows for incremental updates to the prompts, optimizing the object first, then the state, and vice versa. Meanwhile, the optimization of state and object is independent, thus clearer features can be learned to further alleviate the issue of entangling misleading optimization. Moreover, we quantify and analyze the entanglement in CZSL and supplement entanglement rebalancing optimization schemes. DRPT surpasses representative state-of-the-art methods on extensive benchmark datasets, demonstrating superiority in both accuracy and efficiency.

Abstract: Panoptic segmentation is one of the most challenging scene parsing tasks, combining the tasks of semantic segmentation and instance segmentation. While much progress has been made, few works focus on the real-time application of panoptic segmentation methods. In this paper, we revisit the recently introduced K-Net architecture. We propose vital changes to the architecture, training, and inference procedure, which massively decrease latency and improve performance. Our resulting RT-K-Net sets a new state-of-the-art performance for real-time panoptic segmentation methods on the Cityscapes dataset and shows promising results on the challenging Mapillary Vistas dataset. On Cityscapes, RT-K-Net reaches 60.2 % PQ with an average inference time of 32 ms for full resolution 1024x2048 pixel images on a single Titan RTX GPU. On Mapillary Vistas, RT-K-Net reaches 33.2 % PQ with an average inference time of 69 ms. Source code is available at https://github.com/markusschoen/RT-K-Net.

Abstract: We introduce DreamPaint, a framework to intelligently inpaint any e-commerce product on any user-provided context image. The context image can be, for example, the user's own image for virtual try-on of clothes from the e-commerce catalog on themselves, the user's room image for virtual try-on of a piece of furniture from the e-commerce catalog in their room, etc. As opposed to previous augmented-reality (AR)-based virtual try-on methods, DreamPaint does not use, nor does it require, 3D modeling of neither the e-commerce product nor the user context. Instead, it directly uses 2D images of the product as available in product catalog database, and a 2D picture of the context, for example taken from the user's phone camera. The method relies on few-shot fine tuning a pre-trained diffusion model with the masked latents (e.g., Masked DreamBooth) of the catalog images per item, whose weights are then loaded on a pre-trained inpainting module that is capable of preserving the characteristics of the context image. DreamPaint allows to preserve both the product image and the context (environment/user) image without requiring text guidance to describe the missing part (product/context). DreamPaint also allows to intelligently infer the best 3D angle of the product to place at the desired location on the user context, even if that angle was previously unseen in the product's reference 2D images. We compare our results against both text-guided and image-guided inpainting modules and show that DreamPaint yields superior performance in both subjective human study and quantitative metrics.

Abstract: Weakly supervised semantic segmentation with weak labels is a long-lived ill-posed problem. Mainstream methods mainly focus on improving the quality of pseudo labels. In this report, we attempt to explore the potential of 'prompt to masks' from the powerful class-agnostic large segmentation model, segment-anything. Specifically, different weak labels are used as prompts to the segment-anything model, generating precise class masks. The class masks are utilized to generate pseudo labels to train the segmentation networks. We have conducted extensive experiments on PASCAL VOC 2012 dataset. Experiments demonstrate that segment-anything can serve as a good pseudo-label generator. The code will be made publicly available.

Abstract: While developing a new vision-language LLM (VL-LLM) by pre-training on tremendous image-text pairs from scratch can be exceedingly resource-consuming, connecting an existing LLM with a comparatively lightweight visual prompt generator (VPG) becomes a feasible paradigm. However, further tuning the VPG part of the VL-LLM still suffers from indispensable computational costs, i.e., requiring thousands of GPU hours and millions of training data. One alternative solution is to transfer an existing VPG from any existing VL-LLMs for the target VL-LLM. In this work, we for the first time investigate the VPG transferability across LLMs, and explore a solution to reduce the cost of VPG transfer. We first study the VPG transfer across different LLM sizes (e.g., small-to-large), and across different LLM types, through which we diagnose the key factors to maximize the transfer efficiency. Based on our observation, we design a two-stage transfer framework named VPGTrans, which is simple yet highly effective. Through extensive experiments, we demonstrate that VPGTrans helps significantly speed up the transfer learning process without compromising performance. Remarkably, it helps achieve the VPG transfer from BLIP-2 OPT$_\text{2.7B}$ to BLIP-2 OPT$_\text{6.7B}$ with over 10 times speed-up and 10.7% training data compared with connecting a VPG to OPT$_\text{6.7B}$ from scratch. Further, a series of intriguing findings and potential rationales behind them are provided and discussed. Finally, we showcase the practical value of our VPGTrans approach, by customizing two novel VL-LLMs, including VL-LLaMA and VL-Vicuna, with recently released LLaMA and Vicuna LLMs.

Abstract: Extensive work has demonstrated the effectiveness of Vision Transformers. The plain Vision Transformer tends to obtain multi-scale features by selecting fixed layers, or the last layer of features aiming to achieve higher performance in dense prediction tasks. However, this selection is often based on manual operation. And different samples often exhibit different features at different layers (e.g., edge, structure, texture, detail, etc.). This requires us to seek a dynamic adaptive fusion method to filter different layer features. In this paper, unlike previous encoder and decoder work, we design a neck network for adaptive fusion and feature selection, called ViTController. We validate the effectiveness of our method on different datasets and models and surpass previous state-of-the-art methods. Finally, our method can also be used as a plug-in module and inserted into different networks.

Abstract: Recently Transformer has shown good performance in several vision tasks due to its powerful modeling capabilities. To reduce the quadratic complexity caused by the attention, some outstanding work restricts attention to local regions or extends axial interactions. However, these methos often lack the interaction of local and global information, balancing coarse and fine-grained information. To address this problem, we propose AxWin Attention, which models context information in both local windows and axial views. Based on the AxWin Attention, we develop a context-aware vision transformer backbone, named AxWin Transformer, which outperforming the state-of-the-art methods in both classification and downstream segmentation and detection tasks.

Abstract: Face Recognition System (FRS) are shown to be vulnerable to morphed images of newborns. Detecting morphing attacks stemming from face images of newborn is important to avoid unwanted consequences, both for security and society. In this paper, we present a new reference-based/Differential Morphing Attack Detection (MAD) method to detect newborn morphing images using Wavelet Scattering Network (WSN). We propose a two-layer WSN with 250 $\times$ 250 pixels and six rotations of wavelets per layer, resulting in 577 paths. The proposed approach is validated on a dataset of 852 bona fide images and 2460 morphing images constructed using face images of 42 unique newborns. The obtained results indicate a gain of over 10\% in detection accuracy over other existing D-MAD techniques.

Abstract: Recent research has shown that Deep Neural Networks (DNNs) are highly vulnerable to adversarial samples, which are highly transferable and can be used to attack other unknown black-box models. To improve the transferability of adversarial samples, several feature-based adversarial attack methods have been proposed to disrupt neuron activation in middle layers. However, current state-of-the-art feature-based attack methods typically require additional computation costs for estimating the importance of neurons. To address this challenge, we propose a Singular Value Decomposition (SVD)-based feature-level attack method. Our approach is inspired by the discovery that eigenvectors associated with the larger singular values decomposed from the middle layer features exhibit superior generalization and attention properties. Specifically, we conduct the attack by retaining the decomposed Top-1 singular value-associated feature for computing the output logits, which are then combined with the original logits to optimize adversarial perturbations. Our extensive experimental results verify the effectiveness of our proposed method, which significantly enhances the transferability of adversarial samples against various baseline models and defense strategies.The source code of this study is available at \href{https://anonymous.4open.science/r/SVD-SSA-13BF/README.md}.

Abstract: Crude oil is an integral component of the modern world economy. With the growing demand for crude oil due to its widespread applications, accidental oil spills are unavoidable. Even though oil spills are in and themselves difficult to clean up, the first and foremost challenge is to detect spills. In this research, the authors test the feasibility of deep encoder-decoder models that can be trained effectively to detect oil spills. The work compares the results from several segmentation models on high dimensional satellite Synthetic Aperture Radar (SAR) image data. Multiple combinations of models are used in running the experiments. The best-performing model is the one with the ResNet-50 encoder and DeepLabV3+ decoder. It achieves a mean Intersection over Union (IoU) of 64.868% and a class IoU of 61.549% for the "oil spill" class when compared with the current benchmark model, which achieved a mean IoU of 65.05% and a class IoU of 53.38% for the "oil spill" class.

Abstract: Video text spotting refers to localizing, recognizing, and tracking textual elements such as captions, logos, license plates, signs, and other forms of text within consecutive video frames. However, current datasets available for this task rely on quadrilateral ground truth annotations, which may result in including excessive background content and inaccurate text boundaries. Furthermore, methods trained on these datasets often produce prediction results in the form of quadrilateral boxes, which limits their ability to handle complex scenarios such as dense or curved text. To address these issues, we propose a scalable mask annotation pipeline called SAMText for video text spotting. SAMText leverages the SAM model to generate mask annotations for scene text images or video frames at scale. Using SAMText, we have created a large-scale dataset, SAMText-9M, that contains over 2,400 video clips sourced from existing datasets and over 9 million mask annotations. We have also conducted a thorough statistical analysis of the generated masks and their quality, identifying several research topics that could be further explored based on this dataset. The code and dataset will be released at \url{https://github.com/ViTAE-Transformer/SAMText}.

Abstract: In this paper, we introduce a framework ARBEx, a novel attentive feature extraction framework driven by Vision Transformer with reliability balancing to cope against poor class distributions, bias, and uncertainty in the facial expression learning (FEL) task. We reinforce several data pre-processing and refinement methods along with a window-based cross-attention ViT to squeeze the best of the data. We also employ learnable anchor points in the embedding space with label distributions and multi-head self-attention mechanism to optimize performance against weak predictions with reliability balancing, which is a strategy that leverages anchor points, attention scores, and confidence values to enhance the resilience of label predictions. To ensure correct label classification and improve the models' discriminative power, we introduce anchor loss, which encourages large margins between anchor points. Additionally, the multi-head self-attention mechanism, which is also trainable, plays an integral role in identifying accurate labels. This approach provides critical elements for improving the reliability of predictions and has a substantial positive effect on final prediction capabilities. Our adaptive model can be integrated with any deep neural network to forestall challenges in various recognition tasks. Our strategy outperforms current state-of-the-art methodologies, according to extensive experiments conducted in a variety of contexts.

Abstract: Recent progress of deep learning has empowered various intelligent transportation applications, especially in car-sharing platforms. While the traditional operations of the car-sharing service highly relied on human engagements in fleet management, modern car-sharing platforms let users upload car images before and after their use to inspect the cars without a physical visit. To automate the aforementioned inspection task, prior approaches utilized deep neural networks. They commonly employed pre-training, a de-facto technique to establish an effective model under the limited number of labeled datasets. As candidate practitioners who deal with car images would presumably get suffered from the lack of a labeled dataset, we analyzed a sophisticated analogy into the effectiveness of pre-training is important. However, prior studies primarily shed a little spotlight on the effectiveness of pre-training. Motivated by the aforementioned lack of analysis, our study proposes a series of analyses to unveil the effectiveness of various pre-training methods in image recognition tasks at the car-sharing platform. We set two real-world image recognition tasks in the car-sharing platform in a live service, established them under the many-shot and few-shot problem settings, and scrutinized which pre-training method accomplishes the most effective performance in which setting. Furthermore, we analyzed how does the pre-training and fine-tuning convey different knowledge to the neural networks for a precise understanding.

Abstract: The ability to collect a large dataset of human preferences from text-to-image users is usually limited to companies, making such datasets inaccessible to the public. To address this issue, we create a web app that enables text-to-image users to generate images and specify their preferences. Using this web app we build Pick-a-Pic, a large, open dataset of text-to-image prompts and real users' preferences over generated images. We leverage this dataset to train a CLIP-based scoring function, PickScore, which exhibits superhuman performance on the task of predicting human preferences. Then, we test PickScore's ability to perform model evaluation and observe that it correlates better with human rankings than other automatic evaluation metrics. Therefore, we recommend using PickScore for evaluating future text-to-image generation models, and using Pick-a-Pic prompts as a more relevant dataset than MS-COCO. Finally, we demonstrate how PickScore can enhance existing text-to-image models via ranking.

Abstract: The Segment Anything Model (SAM) has demonstrated exceptional performance and versatility, making it a promising tool for various related tasks. In this report, we explore the application of SAM in Weakly-Supervised Semantic Segmentation (WSSS). Particularly, we adapt SAM as the pseudo-label generation pipeline given only the image-level class labels. While we observed impressive results in most cases, we also identify certain limitations. Our study includes performance evaluations on PASCAL VOC and MS-COCO, where we achieved remarkable improvements over the latest state-of-the-art methods on both datasets. We anticipate that this report encourages further explorations of adopting SAM in WSSS, as well as wider real-world applications.

Abstract: With the proliferation of algorithmic decision-making, increased scrutiny has been placed on these systems. This paper explores the relationship between the quality of the training data and the overall fairness of the models trained with such data in the context of supervised classification. We measure key fairness metrics across a range of algorithms over multiple image classification datasets that have a varying level of noise in both the labels and the training data itself. We describe noise in the labels as inaccuracies in the labelling of the data in the training set and noise in the data as distortions in the data, also in the training set. By adding noise to the original datasets, we can explore the relationship between the quality of the training data and the fairness of the output of the models trained on that data.

Abstract: Human and environment sensing are two important topics in Computer Vision and Graphics. Human motion is often captured by inertial sensors, while the environment is mostly reconstructed using cameras. We integrate the two techniques together in EgoLocate, a system that simultaneously performs human motion capture (mocap), localization, and mapping in real time from sparse body-mounted sensors, including 6 inertial measurement units (IMUs) and a monocular phone camera. On one hand, inertial mocap suffers from large translation drift due to the lack of the global positioning signal. EgoLocate leverages image-based simultaneous localization and mapping (SLAM) techniques to locate the human in the reconstructed scene. On the other hand, SLAM often fails when the visual feature is poor. EgoLocate involves inertial mocap to provide a strong prior for the camera motion. Experiments show that localization, a key challenge for both two fields, is largely improved by our technique, compared with the state of the art of the two fields. Our codes are available for research at https://xinyu-yi.github.io/EgoLocate/.

Abstract: Multi-color holograms rely on simultaneous illumination from multiple light sources. These multi-color holograms could utilize light sources better than conventional single-color holograms and can improve the dynamic range of holographic displays. In this letter, we introduce \projectname, the first learned method for estimating the optimal light source powers required for illuminating multi-color holograms. For this purpose, we establish the first multi-color hologram dataset using synthetic images and their depth information. We generate these synthetic images using a trending pipeline combining generative, large language, and monocular depth estimation models. Finally, we train our learned model using our dataset and experimentally demonstrate that \projectname significantly decreases the number of steps required to optimize multi-color holograms from $>1000$ to $70$ iteration steps without compromising image quality.

Abstract: We propose a new dataset and a novel approach to learning hand-object interaction priors for hand and articulated object pose estimation. We first collect a dataset using visual teleoperation, where the human operator can directly play within a physical simulator to manipulate the articulated objects. We record the data and obtain free and accurate annotations on object poses and contact information from the simulator. Our system only requires an iPhone to record human hand motion, which can be easily scaled up and largely lower the costs of data and annotation collection. With this data, we learn 3D interaction priors including a discriminator (in a GAN) capturing the distribution of how object parts are arranged, and a diffusion model which generates the contact regions on articulated objects, guiding the hand pose estimation. Such structural and contact priors can easily transfer to real-world data with barely any domain gap. By using our data and learned priors, our method significantly improves the performance on joint hand and articulated object poses estimation over the existing state-of-the-art methods. The project is available at https://zehaozhu.github.io/ContactArt/ .

Abstract: In this paper, we present TMR, a simple yet effective approach for text to 3D human motion retrieval. While previous work has only treated retrieval as a proxy evaluation metric, we tackle it as a standalone task. Our method extends the state-of-the-art text-to-motion synthesis model TEMOS, and incorporates a contrastive loss to better structure the cross-modal latent space. We show that maintaining the motion generation loss, along with the contrastive training, is crucial to obtain good performance. We introduce a benchmark for evaluation and provide an in-depth analysis by reporting results on several protocols. Our extensive experiments on the KIT-ML and HumanML3D datasets show that TMR outperforms the prior work by a significant margin, for example reducing the median rank from 54 to 19. Finally, we showcase the potential of our approach on moment retrieval. Our code and models are publicly available.

Abstract: We present Neural Fields for LiDAR (NFL), a method to optimise a neural field scene representation from LiDAR measurements, with the goal of synthesizing realistic LiDAR scans from novel viewpoints. NFL combines the rendering power of neural fields with a detailed, physically motivated model of the LiDAR sensing process, thus enabling it to accurately reproduce key sensor behaviors like beam divergence, secondary returns, and ray dropping. We evaluate NFL on synthetic and real LiDAR scans and show that it outperforms explicit reconstruct-then-simulate methods as well as other NeRF-style methods on LiDAR novel view synthesis task. Moreover, we show that the improved realism of the synthesized views narrows the domain gap to real scans and translates to better registration and semantic segmentation performance.

Abstract: Text-to-image models (T2I) offer a new level of flexibility by allowing users to guide the creative process through natural language. However, personalizing these models to align with user-provided visual concepts remains a challenging problem. The task of T2I personalization poses multiple hard challenges, such as maintaining high visual fidelity while allowing creative control, combining multiple personalized concepts in a single image, and keeping a small model size. We present Perfusion, a T2I personalization method that addresses these challenges using dynamic rank-1 updates to the underlying T2I model. Perfusion avoids overfitting by introducing a new mechanism that "locks" new concepts' cross-attention Keys to their superordinate category. Additionally, we develop a gated rank-1 approach that enables us to control the influence of a learned concept during inference time and to combine multiple concepts. This allows runtime-efficient balancing of visual-fidelity and textual-alignment with a single 100KB trained model, which is five orders of magnitude smaller than the current state of the art. Moreover, it can span different operating points across the Pareto front without additional training. Finally, we show that Perfusion outperforms strong baselines in both qualitative and quantitative terms. Importantly, key-locking leads to novel results compared to traditional approaches, allowing to portray personalized object interactions in unprecedented ways, even in one-shot settings.

Abstract: Dataset Distillation aims to distill an entire dataset's knowledge into a few synthetic images. The idea is to synthesize a small number of synthetic data points that, when given to a learning algorithm as training data, result in a model approximating one trained on the original data. Despite recent progress in the field, existing dataset distillation methods fail to generalize to new architectures and scale to high-resolution datasets. To overcome the above issues, we propose to use the learned prior from pre-trained deep generative models to synthesize the distilled data. To achieve this, we present a new optimization algorithm that distills a large number of images into a few intermediate feature vectors in the generative model's latent space. Our method augments existing techniques, significantly improving cross-architecture generalization in all settings.

Abstract: The relatively hot temperature of the human body causes people to turn into long-wave infrared light sources. Since this emitted light has a larger wavelength than visible light, many surfaces in typical scenes act as infrared mirrors with strong specular reflections. We exploit the thermal reflections of a person onto objects in order to locate their position and reconstruct their pose, even if they are not visible to a normal camera. We propose an analysis-by-synthesis framework that jointly models the objects, people, and their thermal reflections, which allows us to combine generative models with differentiable rendering of reflections. Quantitative and qualitative experiments show our approach works in highly challenging cases, such as with curved mirrors or when the person is completely unseen by a normal camera.