Computer Vision and Pattern Recognition (cs.CV)

Abstract: Segmentation of curvilinear structures such as vasculature and road networks is challenging due to relatively weak signals and complex geometry/topology. To facilitate and accelerate large scale annotation, one has to adopt semi-automatic approaches such as proofreading by experts. In this work, we focus on uncertainty estimation for such tasks, so that highly uncertain, and thus error-prone structures can be identified for human annotators to verify. Unlike most existing works, which provide pixel-wise uncertainty maps, we stipulate it is crucial to estimate uncertainty in the units of topological structures, e.g., small pieces of connections and branches. To achieve this, we leverage tools from topological data analysis, specifically discrete Morse theory (DMT), to first capture the structures, and then reason about their uncertainties. To model the uncertainty, we (1) propose a joint prediction model that estimates the uncertainty of a structure while taking the neighboring structures into consideration (inter-structural uncertainty); (2) propose a novel Probabilistic DMT to model the inherent uncertainty within each structure (intra-structural uncertainty) by sampling its representations via a perturb-and-walk scheme. On various 2D and 3D datasets, our method produces better structure-wise uncertainty maps compared to existing works.

Abstract: On the one hand, the dehazing task is an illposedness problem, which means that no unique solution exists. On the other hand, the dehazing task should take into account the subjective factor, which is to give the user selectable dehazed images rather than a single result. Therefore, this paper proposes a multi-output dehazing network by introducing illumination controllable ability, called IC-Dehazing. The proposed IC-Dehazing can change the illumination intensity by adjusting the factor of the illumination controllable module, which is realized based on the interpretable Retinex theory. Moreover, the backbone dehazing network of IC-Dehazing consists of a Transformer with double decoders for high-quality image restoration. Further, the prior-based loss function and unsupervised training strategy enable IC-Dehazing to complete the parameter learning process without the need for paired data. To demonstrate the effectiveness of the proposed IC-Dehazing, quantitative and qualitative experiments are conducted on image dehazing, semantic segmentation, and object detection tasks. Code is available at https://github.com/Xiaofeng-life/ICDehazing.

Abstract: Depth estimation is an important task in various robotics systems and applications. In mobile robotics systems, monocular depth estimation is desirable since a single RGB camera can be deployable at a low cost and compact size. Due to its significant and growing needs, many lightweight monocular depth estimation networks have been proposed for mobile robotics systems. While most lightweight monocular depth estimation methods have been developed using convolution neural networks, the Transformer has been gradually utilized in monocular depth estimation recently. However, massive parameters and large computational costs in the Transformer disturb the deployment to embedded devices. In this paper, we present a Token-Sharing Transformer (TST), an architecture using the Transformer for monocular depth estimation, optimized especially in embedded devices. The proposed TST utilizes global token sharing, which enables the model to obtain an accurate depth prediction with high throughput in embedded devices. Experimental results show that TST outperforms the existing lightweight monocular depth estimation methods. On the NYU Depth v2 dataset, TST can deliver depth maps up to 63.4 FPS in NVIDIA Jetson nano and 142.6 FPS in NVIDIA Jetson TX2, with lower errors than the existing methods. Furthermore, TST achieves real-time depth estimation of high-resolution images on Jetson TX2 with competitive results.

Abstract: The Transformer structures have been widely used in computer vision and have recently made an impact in the area of medical image registration. However, the use of Transformer in most registration networks is straightforward. These networks often merely use the attention mechanism to boost the feature learning as the segmentation networks do, but do not sufficiently design to be adapted for the registration task. In this paper, we propose a novel motion decomposition Transformer (ModeT) to explicitly model multiple motion modalities by fully exploiting the intrinsic capability of the Transformer structure for deformation estimation. The proposed ModeT naturally transforms the multi-head neighborhood attention relationship into the multi-coordinate relationship to model multiple motion modes. Then the competitive weighting module (CWM) fuses multiple deformation sub-fields to generate the resulting deformation field. Extensive experiments on two public brain magnetic resonance imaging (MRI) datasets show that our method outperforms current state-of-the-art registration networks and Transformers, demonstrating the potential of our ModeT for the challenging non-rigid deformation estimation problem. The benchmarks and our code are publicly available at https://github.com/ZAX130/SmileCode.

Abstract: Research in scene graph generation (SGG) usually considers two-stage models, that is, detecting a set of entities, followed by combining them and labeling all possible relationships. While showing promising results, the pipeline structure induces large parameter and computation overhead, and typically hinders end-to-end optimizations. To address this, recent research attempts to train single-stage models that are computationally efficient. With the advent of DETR, a set based detection model, one-stage models attempt to predict a set of subject-predicate-object triplets directly in a single shot. However, SGG is inherently a multi-task learning problem that requires modeling entity and predicate distributions simultaneously. In this paper, we propose Transformers with conditional queries for SGG, namely, TraCQ with a new formulation for SGG that avoids the multi-task learning problem and the combinatorial entity pair distribution. We employ a DETR-based encoder-decoder design and leverage conditional queries to significantly reduce the entity label space as well, which leads to 20% fewer parameters compared to state-of-the-art single-stage models. Experimental results show that TraCQ not only outperforms existing single-stage scene graph generation methods, it also beats many state-of-the-art two-stage methods on the Visual Genome dataset, yet is capable of end-to-end training and faster inference.

Abstract: Recent advancements in neural network-based optical flow estimation often come with prohibitively high computational and memory requirements, presenting challenges in their model adaptation for mobile and low-power use cases. In this paper, we introduce a lightweight low-latency and memory-efficient model, Dynamic Iterative Field Transforms (DIFT), for optical flow estimation feasible for edge applications such as mobile, XR, micro UAVs, robotics and cameras. DIFT follows an iterative refinement framework leveraging variable resolution of cost volumes for correspondence estimation. We propose a memory efficient solution for cost volume processing to reduce peak memory. Also, we present a novel dynamic coarse-to-fine cost volume processing during various stages of refinement to avoid multiple levels of cost volumes. We demonstrate first real-time cost-volume based optical flow DL architecture on Snapdragon 8 Gen 1 HTP efficient mobile AI accelerator with 32 inf/sec and 5.89 EPE (endpoint error) on KITTI with manageable accuracy-performance tradeoffs.

Abstract: Image compression aims to reduce the information redundancy in images. Most existing neural image compression methods rely on side information from hyperprior or context models to eliminate spatial redundancy, but rarely address the channel redundancy. Inspired by the mask sampling modeling in recent self-supervised learning methods for natural language processing and high-level vision, we propose a novel pretraining strategy for neural image compression. Specifically, Cube Mask Sampling Module (CMSM) is proposed to apply both spatial and channel mask sampling modeling to image compression in the pre-training stage. Moreover, to further reduce channel redundancy, we propose the Learnable Channel Mask Module (LCMM) and the Learnable Channel Completion Module (LCCM). Our plug-and-play CMSM, LCMM, LCCM modules can apply to both CNN-based and Transformer-based architectures, significantly reduce the computational cost, and improve the quality of images. Experiments on the public Kodak and Tecnick datasets demonstrate that our method achieves competitive performance with lower computational complexity compared to state-of-the-art image compression methods.

Abstract: This article discusses the opportunities, applications and future directions of large-scale pre-trained models, i.e., foundation models, for analyzing medical images. Medical foundation models have immense potential in solving a wide range of downstream tasks, as they can help to accelerate the development of accurate and robust models, reduce the large amounts of required labeled data, preserve the privacy and confidentiality of patient data. Specifically, we illustrate the "spectrum" of medical foundation models, ranging from general vision models, modality-specific models, to organ/task-specific models, highlighting their challenges, opportunities and applications. We also discuss how foundation models can be leveraged in downstream medical tasks to enhance the accuracy and efficiency of medical image analysis, leading to more precise diagnosis and treatment decisions.

Abstract: Domain adaptive object detection (DAOD) aims to generalize detectors trained on an annotated source domain to an unlabelled target domain. However, existing methods focus on reducing the domain bias of the detection backbone by inferring a discriminative visual encoder, while ignoring the domain bias in the detection head. Inspired by the high generalization of vision-language models (VLMs), applying a VLM as the robust detection backbone following a domain-aware detection head is a reasonable way to learn the discriminative detector for each domain, rather than reducing the domain bias in traditional methods. To achieve the above issue, we thus propose a novel DAOD framework named Domain-Aware detection head with Prompt tuning (DA-Pro), which applies the learnable domain-adaptive prompt to generate the dynamic detection head for each domain. Formally, the domain-adaptive prompt consists of the domain-invariant tokens, domain-specific tokens, and the domain-related textual description along with the class label. Furthermore, two constraints between the source and target domains are applied to ensure that the domain-adaptive prompt can capture the domains-shared and domain-specific knowledge. A prompt ensemble strategy is also proposed to reduce the effect of prompt disturbance. Comprehensive experiments over multiple cross-domain adaptation tasks demonstrate that using the domain-adaptive prompt can produce an effectively domain-related detection head for boosting domain-adaptive object detection.

Abstract: Latent diffusion models (LDMs) exhibit an impressive ability to produce realistic images, yet the inner workings of these models remain mysterious. Even when trained purely on images without explicit depth information, they typically output coherent pictures of 3D scenes. In this work, we investigate a basic interpretability question: does an LDM create and use an internal representation of simple scene geometry? Using linear probes, we find evidence that the internal activations of the LDM encode linear representations of both 3D depth data and a salient-object / background distinction. These representations appear surprisingly early in the denoising process$-$well before a human can easily make sense of the noisy images. Intervention experiments further indicate these representations play a causal role in image synthesis, and may be used for simple high-level editing of an LDM's output.

Abstract: Recently, there has been a growing interest in research concerning document image analysis and recognition in photographic scenarios. However, the lack of labeled datasets for this emerging challenge poses a significant obstacle, as manual annotation can be time-consuming and impractical. To tackle this issue, we present DocAligner, a novel method that streamlines the manual annotation process to a simple step of taking pictures. DocAligner achieves this by establishing dense correspondence between photographic document images and their clean counterparts. It enables the automatic transfer of existing annotations in clean document images to photographic ones and helps to automatically acquire labels that are unavailable through manual labeling. Considering the distinctive characteristics of document images, DocAligner incorporates several innovative features. First, we propose a non-rigid pre-alignment technique based on the document's edges, which effectively eliminates interference caused by significant global shifts and repetitive patterns present in document images. Second, to handle large shifts and ensure high accuracy, we introduce a hierarchical aligning approach that combines global and local correlation layers. Furthermore, considering the importance of fine-grained elements in document images, we present a details recurrent refinement module to enhance the output in a high-resolution space. To train DocAligner, we construct a synthetic dataset and introduce a self-supervised learning approach to enhance its robustness for real-world data. Through extensive experiments, we demonstrate the effectiveness of DocAligner and the acquired dataset. Datasets and codes will be publicly available.

Abstract: This technical report presents our solution to Ball Action Spotting in videos. Our method reached second place in the CVPR'23 SoccerNet Challenge. Details of this challenge can be found at https://www.soccer-net.org/tasks/ball-action-spotting. Our approach is developed based on a baseline model termed E2E-Spot, which was provided by the organizer of this competition. We first generated several variants of the E2E-Spot model, resulting in a candidate model set. We then proposed a strategy for selecting appropriate model members from this set and assigning an appropriate weight to each model. The aim of this strategy is to boost the performance of the resulting model ensemble. Therefore, we call our approach Boosted Model Ensembling (BME). Our code is available at https://github.com/ZJLAB-AMMI/E2E-Spot-MBS.

Abstract: Multi-person pose tracking is an important element for many applications and requires to estimate the human poses of all persons in a video and to track them over time. The association of poses across frames remains an open research problem, in particular for online tracking methods, due to motion blur, crowded scenes and occlusions. To tackle the association challenge, we propose a Dual-Source Attention Transformer that incorporates three core aspects: i) In order to re-identify persons that have been occluded, we propose a pose-conditioned re-identification network that provides an initial embedding and allows to match persons even if the number of visible joints differs between the frames. ii) We incorporate edge embeddings based on temporal pose similarity and the impact of appearance and pose similarity is automatically adapted. iii) We propose an attention based matching layer for pose-to-track association and duplicate removal. We evaluate our approach on Market1501, PoseTrack 2018 and PoseTrack21.

Abstract: We present a novel freehand sketch beautification method, which takes as input a freely drawn sketch of a man-made object and automatically beautifies it both geometrically and structurally. Beautifying a sketch is challenging because of its highly abstract and heavily diverse drawing manner. Existing methods are usually confined to the distribution of their limited training samples and thus cannot beautify freely drawn sketches with rich variations. To address this challenge, we adopt a divide-and-combine strategy. Specifically, we first parse an input sketch into semantic components, beautify individual components by a learned part beautification module based on part-level implicit manifolds, and then reassemble the beautified components through a structure beautification module. With this strategy, our method can go beyond the training samples and handle novel freehand sketches. We demonstrate the effectiveness of our system with extensive experiments and a perceptive study.

Abstract: As the use of collaborative robots (cobots) in industrial manufacturing continues to grow, human action recognition for effective human-robot collaboration becomes increasingly important. This ability is crucial for cobots to act autonomously and assist in assembly tasks. Recently, skeleton-based approaches are often used as they tend to generalize better to different people and environments. However, when processing skeletons alone, information about the objects a human interacts with is lost. Therefore, we present a novel approach of integrating object information into skeleton-based action recognition. We enhance two state-of-the-art methods by treating object centers as further skeleton joints. Our experiments on the assembly dataset IKEA ASM show that our approach improves the performance of these state-of-the-art methods to a large extent when combining skeleton joints with objects predicted by a state-of-the-art instance segmentation model. Our research sheds light on the benefits of combining skeleton joints with object information for human action recognition in assembly tasks. We analyze the effect of the object detector on the combination for action classification and discuss the important factors that must be taken into account.

Abstract: Pose and motion priors are crucial for recovering realistic and accurate human motion from noisy observations. Substantial progress has been made on pose and shape estimation from images, and recent works showed impressive results using priors to refine frame-wise predictions. However, a lot of motion priors only model transitions between consecutive poses and are used in time-consuming optimization procedures, which is problematic for many applications requiring real-time motion capture. We introduce Motion-DVAE, a motion prior to capture the short-term dependencies of human motion. As part of the dynamical variational autoencoder (DVAE) models family, Motion-DVAE combines the generative capability of VAE models and the temporal modeling of recurrent architectures. Together with Motion-DVAE, we introduce an unsupervised learned denoising method unifying regression- and optimization-based approaches in a single framework for real-time 3D human pose estimation. Experiments show that the proposed approach reaches competitive performance with state-of-the-art methods while being much faster.

Abstract: Generating realistic human 3D reconstructions using image or video data is essential for various communication and entertainment applications. While existing methods achieved impressive results for body and facial regions, realistic hair modeling still remains challenging due to its high mechanical complexity. This work proposes an approach capable of accurate hair geometry reconstruction at a strand level from a monocular video or multi-view images captured in uncontrolled lighting conditions. Our method has two stages, with the first stage performing joint reconstruction of coarse hair and bust shapes and hair orientation using implicit volumetric representations. The second stage then estimates a strand-level hair reconstruction by reconciling in a single optimization process the coarse volumetric constraints with hair strand and hairstyle priors learned from the synthetic data. To further increase the reconstruction fidelity, we incorporate image-based losses into the fitting process using a new differentiable renderer. The combined system, named Neural Haircut, achieves high realism and personalization of the reconstructed hairstyles.

Abstract: 3D multi-object tracking (MOT) is vital for many applications including autonomous driving vehicles and service robots. With the commonly used tracking-by-detection paradigm, 3D MOT has made important progress in recent years. However, these methods only use the detection boxes of the current frame to obtain trajectory-box association results, which makes it impossible for the tracker to recover objects missed by the detector. In this paper, we present TrajectoryFormer, a novel point-cloud-based 3D MOT framework. To recover the missed object by detector, we generates multiple trajectory hypotheses with hybrid candidate boxes, including temporally predicted boxes and current-frame detection boxes, for trajectory-box association. The predicted boxes can propagate object's history trajectory information to the current frame and thus the network can tolerate short-term miss detection of the tracked objects. We combine long-term object motion feature and short-term object appearance feature to create per-hypothesis feature embedding, which reduces the computational overhead for spatial-temporal encoding. Additionally, we introduce a Global-Local Interaction Module to conduct information interaction among all hypotheses and models their spatial relations, leading to accurate estimation of hypotheses. Our TrajectoryFormer achieves state-of-the-art performance on the Waymo 3D MOT benchmarks.

Abstract: In the process of product design and digital fabrication, the structural analysis of a designed prototype is a fundamental and essential step. However, such a step is usually invisible or inaccessible to designers at the early sketching phase. This limits the user's ability to consider a shape's physical properties and structural soundness. To bridge this gap, we introduce a novel approach Sketch2Stress that allows users to perform structural analysis of desired objects at the sketching stage. This method takes as input a 2D freehand sketch and one or multiple locations of user-assigned external forces. With the specially-designed two-branch generative-adversarial framework, it automatically predicts a normal map and a corresponding structural stress map distributed over the user-sketched underlying object. In this way, our method empowers designers to easily examine the stress sustained everywhere and identify potential problematic regions of their sketched object. Furthermore, combined with the predicted normal map, users are able to conduct a region-wise structural analysis efficiently by aggregating the stress effects of multiple forces in the same direction. Finally, we demonstrate the effectiveness and practicality of our system with extensive experiments and user studies.

Abstract: We propose a new object-centric video prediction algorithm based on the deep latent particle (DLP) representation. In comparison to existing slot- or patch-based representations, DLPs model the scene using a set of keypoints with learned parameters for properties such as position and size, and are both efficient and interpretable. Our method, deep dynamic latent particles (DDLP), yields state-of-the-art object-centric video prediction results on several challenging datasets. The interpretable nature of DDLP allows us to perform ``what-if'' generation -- predict the consequence of changing properties of objects in the initial frames, and DLP's compact structure enables efficient diffusion-based unconditional video generation. Videos, code and pre-trained models are available: https://taldatech.github.io/ddlp-web

Abstract: Biological vision systems make adaptive use of context to recognize objects in new settings with novel contexts as well as occluded or blurry objects in familiar settings. In this paper, we investigate how vision models adaptively use context for out-of-distribution (OOD) generalization and leverage our analysis results to improve model OOD generalization. First, we formulate two distinct OOD settings where the contexts are either irrelevant (Background-Invariance) or beneficial (Object-Disambiguation), reflecting the diverse contextual challenges faced in biological vision. We then analyze model performance in these two different OOD settings and demonstrate that models that excel in one setting tend to struggle in the other. Notably, prior works on learning causal features improve on one setting but hurt in the other. This underscores the importance of generalizing across both OOD settings, as this ability is crucial for both human cognition and robust AI systems. Next, to better understand the model properties contributing to OOD generalization, we use representational geometry analysis and our own probing methods to examine a population of models, and we discover that those with more factorized representations and appropriate feature weighting are more successful in handling Background-Invariance and Object-Disambiguation tests. We further validate these findings through causal intervention on representation factorization and feature weighting to demonstrate their causal effect on performance. Lastly, we propose new augmentation methods to enhance model generalization. These methods outperform strong baselines, yielding improvements in both in-distribution and OOD tests. In conclusion, to replicate the generalization abilities of biological vision, computer vision models must have factorized object vs. background representations and appropriately weight both kinds of features.

Abstract: Three-dimensional (3D) point cloud analysis has become one of the attractive subjects in realistic imaging and machine visions due to its simplicity, flexibility and powerful capacity of visualization. Actually, the representation of scenes and buildings using 3D shapes and formats leveraged many applications among which automatic driving, scenes and objects reconstruction, etc. Nevertheless, working with this emerging type of data has been a challenging task for objects representation, scenes recognition, segmentation, and reconstruction. In this regard, a significant effort has recently been devoted to developing novel strategies, using different techniques such as deep learning models. To that end, we present in this paper a comprehensive review of existing tasks on 3D point cloud: a well-defined taxonomy of existing techniques is performed based on the nature of the adopted algorithms, application scenarios, and main objectives. Various tasks performed on 3D point could data are investigated, including objects and scenes detection, recognition, segmentation and reconstruction. In addition, we introduce a list of used datasets, we discuss respective evaluation metrics and we compare the performance of existing solutions to better inform the state-of-the-art and identify their limitations and strengths. Lastly, we elaborate on current challenges facing the subject of technology and future trends attracting considerable interest, which could be a starting point for upcoming research studies

Abstract: Machine learning in medical imaging often faces a fundamental dilemma, namely the small sample size problem. Many recent studies suggest using multi-domain data pooled from different acquisition sites/datasets to improve statistical power. However, medical images from different sites cannot be easily shared to build large datasets for model training due to privacy protection reasons. As a promising solution, federated learning, which enables collaborative training of machine learning models based on data from different sites without cross-site data sharing, has attracted considerable attention recently. In this paper, we conduct a comprehensive survey of the recent development of federated learning methods in medical image analysis. We first introduce the background and motivation of federated learning for dealing with privacy protection and collaborative learning issues in medical imaging. We then present a comprehensive review of recent advances in federated learning methods for medical image analysis. Specifically, existing methods are categorized based on three critical aspects of a federated learning system, including client end, server end, and communication techniques. In each category, we summarize the existing federated learning methods according to specific research problems in medical image analysis and also provide insights into the motivations of different approaches. In addition, we provide a review of existing benchmark medical imaging datasets and software platforms for current federated learning research. We also conduct an experimental study to empirically evaluate typical federated learning methods for medical image analysis. This survey can help to better understand the current research status, challenges and potential research opportunities in this promising research field.

Abstract: In the era of rapid digitalization and artificial intelligence advancements, the development of DeepFake technology has posed significant security and privacy concerns. This paper presents an effective measure to assess the visual realism of DeepFake videos. We utilize an ensemble of two Convolutional Neural Network (CNN) models: Eva and ConvNext. These models have been trained on the DeepFake Game Competition (DFGC) 2022 dataset and aim to predict Mean Opinion Scores (MOS) from DeepFake videos based on features extracted from sequences of frames. Our method secured the third place in the recent DFGC on Visual Realism Assessment held in conjunction with the 2023 International Joint Conference on Biometrics (IJCB 2023). We provide an over\-view of the models, data preprocessing, and training procedures. We also report the performance of our models against the competition's baseline model and discuss the implications of our findings.

Abstract: Self-supervised learning is an effective way for label-free model pre-training, especially in the video domain where labeling is expensive. Existing self-supervised works in the video domain use varying experimental setups to demonstrate their effectiveness and comparison across approaches becomes challenging with no standard benchmark. In this work, we first provide a benchmark that enables a comparison of existing approaches on the same ground. Next, we study five different aspects of self-supervised learning important for videos; 1) dataset size, 2) complexity, 3) data distribution, 4) data noise, and, 5)feature analysis. To facilitate this study, we focus on seven different methods along with seven different network architectures and perform an extensive set of experiments on 5 different datasets with an evaluation of two different downstream tasks. We present several interesting insights from this study which span across different properties of pretraining and target datasets, pretext-tasks, and model architectures among others. We further put some of these insights to the real test and propose an approach that requires a limited amount of training data and outperforms existing state-of-the-art approaches which use 10x pretraining data. We believe this work will pave the way for researchers to a better understanding of self-supervised pretext tasks in video representation learning.

Abstract: Existing offboard 3D detectors always follow a modular pipeline design to take advantage of unlimited sequential point clouds. We have found that the full potential of offboard 3D detectors is not explored mainly due to two reasons: (1) the onboard multi-object tracker cannot generate sufficient complete object trajectories, and (2) the motion state of objects poses an inevitable challenge for the object-centric refining stage in leveraging the long-term temporal context representation. To tackle these problems, we propose a novel paradigm of offboard 3D object detection, named DetZero. Concretely, an offline tracker coupled with a multi-frame detector is proposed to focus on the completeness of generated object tracks. An attention-mechanism refining module is proposed to strengthen contextual information interaction across long-term sequential point clouds for object refining with decomposed regression methods. Extensive experiments on Waymo Open Dataset show our DetZero outperforms all state-of-the-art onboard and offboard 3D detection methods. Notably, DetZero ranks 1st place on Waymo 3D object detection leaderboard with 85.15 mAPH (L2) detection performance. Further experiments validate the application of taking the place of human labels with such high-quality results. Our empirical study leads to rethinking conventions and interesting findings that can guide future research on offboard 3D object detection.

Abstract: Neural Radiance Fields (NeRF) have shown impressive novel view synthesis results; nonetheless, even thorough recordings yield imperfections in reconstructions, for instance due to poorly observed areas or minor lighting changes. Our goal is to mitigate these imperfections from various sources with a joint solution: we take advantage of the ability of generative adversarial networks (GANs) to produce realistic images and use them to enhance realism in 3D scene reconstruction with NeRFs. To this end, we learn the patch distribution of a scene using an adversarial discriminator, which provides feedback to the radiance field reconstruction, thus improving realism in a 3D-consistent fashion. Thereby, rendering artifacts are repaired directly in the underlying 3D representation by imposing multi-view path rendering constraints. In addition, we condition a generator with multi-resolution NeRF renderings which is adversarially trained to further improve rendering quality. We demonstrate that our approach significantly improves rendering quality, e.g., nearly halving LPIPS scores compared to Nerfacto while at the same time improving PSNR by 1.4dB on the advanced indoor scenes of Tanks and Temples.

Abstract: Recent large vision-language models such as CLIP have shown remarkable out-of-distribution (OOD) detection and generalization performance. However, their zero-shot in-distribution (ID) accuracy is often limited for downstream datasets. Recent CLIP-based fine-tuning methods such as prompt learning have demonstrated significant improvements in ID classification and OOD generalization where OOD labels are available. Nonetheless, it remains unclear whether the model is reliable to semantic shifts without OOD labels. In this paper, we aim to bridge the gap and present a comprehensive study to understand how fine-tuning impact OOD detection for few-shot downstream tasks. By framing OOD detection as multi-modal concept matching, we establish a connection between fine-tuning methods and various OOD scores. Our results suggest that a proper choice of OOD scores is essential for CLIP-based fine-tuning. In particular, the maximum concept matching (MCM) score provides a promising solution consistently. We also show that prompt learning demonstrates the state-of-the-art OOD detection performance over the zero-shot counterpart.

Abstract: Deep learning models for image classification have often used a resolution of $224\times224$ pixels for computational reasons. This study investigates the effect of image resolution on chest X-ray classification performance, using the ChestX-ray14 dataset. The results show that a higher image resolution, specifically $1024\times1024$ pixels, has the best overall classification performance, with a slight decline in performance between $256\times256$ to $512\times512$ pixels for most of the pathological classes. Comparison of saliency map-generated bounding boxes revealed that commonly used resolutions are insufficient for finding most pathologies.

Abstract: Flash is an essential tool as it often serves as the sole controllable light source in everyday photography. However, the use of flash is a binary decision at the time a photograph is captured with limited control over its characteristics such as strength or color. In this work, we study the computational control of the flash light in photographs taken with or without flash. We present a physically motivated intrinsic formulation for flash photograph formation and develop flash decomposition and generation methods for flash and no-flash photographs, respectively. We demonstrate that our intrinsic formulation outperforms alternatives in the literature and allows us to computationally control flash in in-the-wild images.

Abstract: Common editing operations performed by professional photographers include the cleanup operations: de-emphasizing distracting elements and enhancing subjects. These edits are challenging, requiring a delicate balance between manipulating the viewer's attention while maintaining photo realism. While recent approaches can boast successful examples of attention attenuation or amplification, most of them also suffer from frequent unrealistic edits. We propose a realism loss for saliency-guided image enhancement to maintain high realism across varying image types, while attenuating distractors and amplifying objects of interest. Evaluations with professional photographers confirm that we achieve the dual objective of realism and effectiveness, and outperform the recent approaches on their own datasets, while requiring a smaller memory footprint and runtime. We thus offer a viable solution for automating image enhancement and photo cleanup operations.

Abstract: Neural Radiance Fields (NeRF) have become an increasingly popular representation to capture high-quality appearance and shape of scenes and objects. However, learning generalizable NeRF priors over categories of scenes or objects has been challenging due to the high dimensionality of network weight space. To address the limitations of existing work on generalization, multi-view consistency and to improve quality, we propose HyP-NeRF, a latent conditioning method for learning generalizable category-level NeRF priors using hypernetworks. Rather than using hypernetworks to estimate only the weights of a NeRF, we estimate both the weights and the multi-resolution hash encodings resulting in significant quality gains. To improve quality even further, we incorporate a denoise and finetune strategy that denoises images rendered from NeRFs estimated by the hypernetwork and finetunes it while retaining multiview consistency. These improvements enable us to use HyP-NeRF as a generalizable prior for multiple downstream tasks including NeRF reconstruction from single-view or cluttered scenes and text-to-NeRF. We provide qualitative comparisons and evaluate HyP-NeRF on three tasks: generalization, compression, and retrieval, demonstrating our state-of-the-art results.

Abstract: Recently, large language models (LLMs) have made significant advancements in natural language understanding and generation. However, their potential in computer vision remains largely unexplored. In this paper, we introduce a new, exploratory approach that enables LLMs to process images using the Scalable Vector Graphics (SVG) format. By leveraging the XML-based textual descriptions of SVG representations instead of raster images, we aim to bridge the gap between the visual and textual modalities, allowing LLMs to directly understand and manipulate images without the need for parameterized visual components. Our method facilitates simple image classification, generation, and in-context learning using only LLM capabilities. We demonstrate the promise of our approach across discriminative and generative tasks, highlighting its (i) robustness against distribution shift, (ii) substantial improvements achieved by tapping into the in-context learning abilities of LLMs, and (iii) image understanding and generation capabilities with human guidance. Our code, data, and models can be found here https://github.com/mu-cai/svg-llm.