Supervised person re-identification (ReID) often has poor scalability and usability in real-world deployments due to domain gaps and the lack of annotations for the target domain data. Unsupervised person ReID through domain adaptation is attractive yet challenging. Existing unsupervised ReID approaches often fail in correctly identifying the positive samples and negative samples through the distance-based matching/ranking. The two distributions of distances for positive sample pairs (Pos-distr) and negative sample pairs (Neg-distr) are often not well separated, having large overlap. To address this problem, we introduce a global distance-distributions separation (GDS) constraint over the two distributions to encourage the clear separation of positive and negative samples from a global view. We model the two global distance distributions as Gaussian distributions and push apart the two distributions while encouraging their sharpness in the unsupervised training process. Particularly, to model the distributions from a global view and facilitate the timely updating of the distributions and the GDS related losses, we leverage a momentum update mechanism for building and maintaining the distribution parameters (mean and variance) and calculate the loss on the fly during the training. Distribution-based hard mining is proposed to further promote the separation of the two distributions. We validate the effectiveness of the GDS constraint in unsupervised ReID networks. Extensive experiments on multiple ReID benchmark datasets show our method leads to significant improvement over the baselines and achieves the state-of-the-art performance.
论文检索
输入标题、作者或关键词，从 1,358 篇学术成果中精准定位
Multi-person pose estimation is challenging because it localizes body keypoints for multiple persons simultaneously. Previous methods can be divided into two streams, \ie top-down and bottom-up methods. The top-down methods localize keypoints after human detection, while the bottom-up methods localize keypoints directly and then cluster/group them for different persons, which are generally more efficient than top-down methods. However, in existing bottom-up methods, the keypoint grouping is usually solved independently from keypoint detection, making them not end-to-end trainable and have sub-optimal performance. In this paper, we investigate a new perspective of human part grouping and reformulate it as a graph clustering task. Especially, we propose a novel differentiable Hierarchical Graph Grouping (HGG) method to learn the graph grouping in bottom-up multi-person pose estimation task. Moreover, HGG is easily embedded into main-stream bottom-up methods. It takes human keypoint candidates as graph nodes and clusters keypoints in a multi-layer graph neural network model. The modules of HGG can be trained end-to-end with the keypoint detection network and is able to supervise the grouping process in a hierarchical manner. To improve the discrimination of the clustering, we add a set of edge discriminators and macro-node discriminators. Extensive experiments on both COCO and OCHuman datasets demonstrate that the proposed method improves the performance of bottom-up pose estimation methods.
Neural architecture search (NAS) methods have shown promising results discovering models that are both accurate and fast. For NAS, training a one-shot model has became a popular strategy to rank the relative quality of different architectures (child models) using a single set of shared weights. However, while one-shot model weights can effectively rank different network architectures, the absolute accuracies from these shared weights are typically far below those obtained from stand-alone training. To compensate, existing methods assume that the weights must be retrained, finetuned, or otherwise post-processed after the search is completed. These steps significantly increase the compute requirements and complexity of the architecture search and model deployment. In this work, we propose BigNAS, an approach that challenges the conventional wisdom that post-processing of the weights is necessary to get good prediction accuracies. Without extra retraining or post-processing steps, we are able to train a single set of shared weights on ImageNet and use these weights to obtain child models whose sizes range from 200 to 1000 MFLOPs. Our discovered model family, BigNASModels, achieve top-1 accuracies ranging from 76.5% to 80.9%, surpassing state-of-the-art models in this range including EfficientNets and Once-for-All networks without extra retraining or post-processing. We present ablative study and analysis to further understand the proposed BigNASModels.
We propose a method to train a model so it can learn new classification tasks while improving with each task solved. This amounts to combining meta-learning with incremental learning. Different tasks can have disjoint classes, so one cannot directly align different classifiers as done in model distillation. On the other hand, simply aligning features shared by all classes does not allow the base model sufficient flexibility to evolve to solve new tasks. We therefore indirectly align features relative to a minimal set of anchor classes. Such indirect discriminant alignment (IDA) adapts a new model to old classes without the need to re-process old data, while leaving maximum flexibility for the model to adapt to new tasks. This process enables incrementally improving the model by processing multiple learning episodes, each representing a different learning task, even with few training examples. Experiments on few-shot learning benchmarks show that this incremental approach performs favorably compared to training the model with the entire dataset at once.
The current spike of hyper-realistic faces artificially generated using deepfakes calls for media forensics solutions that are tailored to video streams and work reliably with a low false alarm rate at the video level. We present a method for deepfake detection based on a two-branch network structure that isolates digitally manipulated faces by learning to amplify artifacts while suppressing the high-level face content. Unlike current methods that extract spatial frequencies as a preprocessing step, we propose a two-branch structure: one branch propagates the original information, while the other branch suppresses the face content yet amplifies multi-band frequencies using a Laplacian of Gaussian (LoG) as a bottleneck layer. To better isolate manipulated faces, we derive a novel cost function that, unlike regular classification, compresses the variability of natural faces and pushes away the unrealistic facial samples in the feature space. Our two novel components show promising results on the FaceForensics++, Celeb-DF, and Facebook's DFDC preview benchmarks, when compared to prior work. We then offer a full, detailed ablation study of our network architecture and cost function. Finally, although the bar is still high to get very remarkable figures at a very low false alarm rate, our study shows that we can achieve good video-level performance when cross-testing in terms of video-level AUC.
We investigate the role of self-supervised learning (SSL) in the context of few-shot learning. Although recent research has shown the benefits of SSL on large unlabeled datasets, its utility on small datasets is relatively unexplored. We find that SSL reduces the relative error rate of few-shot meta-learners by 4%-27%, even when the datasets are small and only utilizing images within the datasets. The improvements are greater when the training set is smaller or the task is more challenging. Although the benefits of SSL may increase with larger training sets, we observe that SSL can hurt the performance when the distributions of images used for meta-learning and SSL are different. We conduct a systematic study by varying the degree of domain shift and analyzing the performance of several meta-learners on a multitude of domains. Based on this analysis we present a technique that automatically selects images for SSL from a large, generic pool of unlabeled images for a given dataset that provides further improvements.
Despite considerable progress, state of the art image captioning models produce generic captions, leaving out important image details. Furthermore, these systems may even misrepresent the image in order to produce a simpler caption consisting of common concepts. In this paper, we first analyze both modern captioning systems and evaluation metrics through empirical experiments to quantify these phenomena. We find that modern captioning systems return higher likelihoods for incorrect distractor sentences compared to ground truth captions, and that evaluation metrics like SPICE can be 'topped' using simple captioning systems relying on object detectors. Inspired by these observations, we design a new metric (SPICE-U) by introducing a notion of uniqueness over the concepts generated in a caption. We show that SPICE-U is better correlated with human judgements compared to SPICE, and effectively captures notions of diversity and descriptiveness. Finally, we also demonstrate a general technique to improve any existing captioning model -- by using mutual information as a re-ranking objective during decoding. Empirically, this results in more unique and informative captions, and improves three different state-of-the-art models on SPICE-U as well as average score over existing metrics.
Efficient representation of articulated objects such as human bodies is an important problem in computer vision and graphics. To efficiently simulate deformation, existing approaches represent 3D objects using polygonal meshes and deform them using skinning techniques. This paper introduces neural articulated shape approximation (NASA), an alternative framework that enables efficient representation of articulated deformable objects using neural indicator functions that are conditioned on pose. Occupancy testing using NASA is straightforward, circumventing the complexity of meshes and the issue of water-tightness. We demonstrate the effectiveness of NASA for 3D tracking applications, and discuss other potential extensions.
Often the best performing deep neural models are ensembles of multiple base-level networks, nevertheless, ensemble learning with respect to domain adaptive person re-ID remains unexplored. In this paper, we propose a multiple expert brainstorming network (MEB-Net) for domain adaptive person re-ID, opening up a promising direction about model ensemble problem under unsupervised conditions. MEB-Net adopts a mutual learning strategy, where multiple networks with different architectures are pre-trained within a source domain as expert models equipped with specific features and knowledge, while the adaptation is then accomplished through brainstorming (mutual learning) among expert models. MEB-Net accommodates the heterogeneity of experts learned with different architectures and enhances discrimination capability of the adapted re-ID model, by introducing a regularization scheme about authority of experts. Extensive experiments on large-scale datasets (Market-1501 and DukeMTMC-reID) demonstrate the superior performance of MEB-Net over the state-of-the-arts.
This work presents a novel First-person View based Trajectory predicting model (FvTraj) to estimate the future trajectories of pedestrians in a scene given their observed trajectories and the corresponding first-person view images. First, we render first-person view images using our in-house built First-person View Simulator (FvSim), given the ground-level 2D trajectories. Then, based on multi-head attention mechanisms, we design a social-aware attention module to model social interactions between pedestrians, and a view-aware attention module to capture the relations between historical motion states and visual features from the first-person view images. Our results show the dynamic scene contexts with ego-motions captured by first-person view images via FvSim are valuable and effective for trajectory prediction. Using this simulated first-person view images, our well structured FvTraj model achieves state-of-the-art performance.
Face anti-spoofing (FAS) plays a vital role in securing the face recognition systems from presentation attacks. Most existing FAS methods capture various cues (e.g., texture, depth and reflection) to distinguish the live faces from the spoofing faces. All these cues are based on the discrepancy among physical materials (e.g., skin, glass, paper and silicone). In this paper we rephrase face anti-spoofing as a material recognition problem and combine it with classical human material perception, intending to extract discriminative and robust features for FAS. To this end, we propose the Bilateral Convolutional Networks (BCN), which is able to capture intrinsic material-based patterns via aggregating multi-level bilateral macro- and micro- information. Furthermore, Multi-level Feature Refinement Module (MFRM) and multi-head supervision are utilized to learn more robust features. Comprehensive experiments are performed on six benchmark datasets, and the proposed method achieves superior performance on both intra- and cross-dataset testings. One highlight is that we achieve overall 11.3$\pm$9.5\% EER for cross-type testing in SiW-M dataset, which significantly outperforms previous results. We hope this work will facilitate future cooperation between FAS and material communities.
We propose a brand new benchmark for analyzing causality in traffic accident videos by decomposing an accident into a pair of events, cause and effect. We collect videos containing traffic accident scenes and annotate cause and effect events for each accident with their temporal intervals and semantic labels; such annotations are not available in existing datasets for accident anticipation task. Our dataset has the following two advantages over the existing ones, which would facilitate practical research for causality analysis. First, the decomposition of an accident into cause and effect events provides atomic cues for reasoning on a complex environment and planning future actions. Second, the prediction of cause and effect in an accident makes a system more interpretable to humans, which mitigates the ambiguity of legal liabilities among agents engaged in the accident. Using the proposed dataset, we analyze accidents by localizing the temporal intervals of their causes and effects and classifying the semantic labels of the accidents. The dataset as well as the implementations of baseline models are available in the code repository.
The field of Few-Shot Learning (FSL), or learning from very few (typically $1$ or $5$) examples per novel class (unseen during training), has received a lot of attention and significant performance advances in the recent literature. While number of techniques have been proposed for FSL, several factors have emerged as most important for FSL performance, awarding SOTA even to the simplest of techniques. These are: the backbone architecture (bigger is better), type of pre-training on the base classes (meta-training vs regular multi-class, currently regular wins), quantity and diversity of the base classes set (the more the merrier, resulting in richer and better adaptive features), and the use of self-supervised tasks during pre-training (serving as a proxy for increasing the diversity of the base set). In this paper we propose yet another simple technique that is important for the few shot learning performance - a search for a compact feature sub-space that is discriminative for a given few-shot test task. We show that the Task-Adaptive Feature Sub-Space Learning (TAFSSL) can significantly boost the performance in FSL scenarios when some additional unlabeled data accompanies the novel few-shot task, be it either the set of unlabeled queries (transductive FSL) or some additional set of unlabeled data samples (semi-supervised FSL). Specifically, we show that on the challenging extit{mini}ImageNet and extit{tiered}ImageNet benchmarks, TAFSSL can improve the current state-of-the-art in both transductive and semi-supervised FSL settings by more than $5\%$, while increasing the benefit of using unlabeled data in FSL to above $10\%$ performance gain.
This paper strives to localize the temporal extent of an action in a long untrimmed video. Where existing work leverages many examples with their start, their ending, and/or the class of the action during training time, we propose few-shot common action localization. The start and end of an action in a long untrimmed video is determined based on just a hand-full of trimmed video examples containing the same action, without knowing their common class label. To address this task, we introduce a new 3D convolutional network architecture able to align representations from the support videos with the relevant query video segments. The network contains: (i) a mutual enhancement module to simultaneously complement the representation of the few trimmed support videos and the untrimmed query video; (ii) a progressive alignment module that iteratively fuses the support videos into the query branch; and (iii) a pairwise matching module to weigh the importance of different support videos. Evaluation of few-shot common action localization in untrimmed videos containing a single or multiple action instances demonstrates the effectiveness and general applicability of our proposal.
In this paper, we propose an end-to-end learnable approach that detects static urban objects from multiple views, re-identifies instances, and finally assigns a geographic position per object. Our method relies on a Graph Neural Network (GNN) to, detect all objects and out-put their geographic positions given images and approximate camera poses as input. Our GNN simultaneously models relative pose and image evidence and is further able to deal with an arbitrary number of input views. Our method is robust to occlusion, with a similar appearance of neighboring objects, and severe changes in viewpoints by jointly reasoning about visual image appearance and relative pose. Experimental evaluation on two challenging, large-scale datasets and comparison with state-of-the-art methods show significant and systematic improvements both in accuracy and efficiency, with 2-6% gain in the detection and re-IDaverage precision as well as 8x reduction of training time.
Long-term camera re-localization is an important task with numerous computer vision and robotics applications. Whilst various outdoor benchmarks exist that target lighting, weather and seasonal changes, far less attention has been paid to appearance changes that occur indoors. This has led to a mismatch between popular indoor benchmarks, which focus on static scenes, and indoor environments that are of interest for many real-world applications. In this paper, we adapt 3RScan -- a recently introduced indoor RGB-D dataset designed for object instance re-localization -- to create RIO10, a new long-term camera re-localization benchmark focused on indoor scenes. We propose new metrics for evaluating camera re-localization and explore how state-of-the-art camera re-localizers perform according to these metrics. We also examine in detail how different types of scene change affect the performance of different methods, based on novel ways of detecting such changes in a given RGB-D frame. Our results clearly show that long-term indoor re-localization is an unsolved problem. Our benchmark and tools are publicly available at waldjohannau.github.io/RIO10
Vision perception is one of the most important components for a computer or robot to understand the surrounding scene and achieve autonomous applications. However, most of the vision models are based on the RGB sensors, which in general are vulnerable to the insufficient lighting condition. In contrast, the depth camera, another widely-used visual sensor, is capable of perceiving 3D information and being more robust to the lack of illumination, but unable to obtain appearance details of the surrounding environment compared to RGB cameras. To make RGB-based vision models workable for the low-lighting scenario, prior methods focus on learning the colorization on depth maps captured by depth cameras, such that the vision models can still achieve reasonable performance on colorized depth maps. However, the colorization produced in this manner is usually unrealistic and constrained to the specific vision model, thus being hard to generalize for other tasks to use. In this paper, we propose a depth map colorization method via disentangling appearance and structure factors, so that our model could 1) learn depth-invariant appearance features from an appearance reference and 2) generate colorized images by combining a given depth map and the appearance feature obtained from any reference. We conduct extensive experiments to show that our colorization results are more realistic and diverse in comparison to several image translation baselines.
The novelty detection models learn a decision boundary around multiple categories of a given dataset. This helps such models in detecting any novel classes encountered during testing. However, in many cases, the test data distribution can be different from that of the training data. For such cases, the novelty detection models risk detecting a known class as novel due to the dataset distribution shift. This scenario is often ignored while working with novelty detection. To this end, we consider the problem of multiple class novelty detection under dataset distribution shift to improve the novelty detection performance. Firstly, we discuss the problem setting in detail and show how it affects the performance of current novelty detection methods. Secondly, we show that one could improve those novelty detection methods with a simple integration of domain adversarial loss. Finally, we propose a method which brings together the techniques from novelty detection and domain adaptation to improve generalization of multiple class novelty detection on different domains. We evaluate the proposed method on digits and object recognition datasets and show that it provides improvements over the baseline methods.
We present an end-to-end 3D reconstruction of a scene by directly regressing a truncated signed distance function (TSDF) from a set of posed RGB images. Traditional approaches to 3D reconstruction rely on an intermediate representation of depth maps prior to estimating a full 3D model of a scene. We hypothesize that a direct regression to 3D is more effective. A 2D CNN extracts features from each image independently which are then back-projected and accumulated into a voxel volume using the camera intrinsics and extrinsics. After accumulation, a 3D CNN refines the accumulated features and predicts the TSDF values. Additionally, semantic segmentation of the 3D model is obtained without significant computation. This approach is evaluated on Scannet dataset where we significantly outperform state-of-the-art baselines (deep multiview stereo followed by traditional TSDF fusion) both quantitatively and qualitatively. We compare our 3D semantic segmentation to prior methods that use a depth sensor since no previous work attempts the problem with only RGB input.
We introduce a new problem of mph{retrieving} 3D models that are mph{deformable} to a given query shape and present a novel deep mph{deformation-aware} embedding to solve this retrieval task. 3D model retrieval is a fundamental operation for recovering a clean and complete 3D model from a noisy and partial 3D scan. However, given a finite collection of 3D shapes, even the closest model to a query may not be satisfactory. This motivates us to apply 3D model deformation techniques to adapt the retrieved model so as to better fit the query. Yet, certain restrictions are enforced in most 3D deformation techniques to preserve important features of the original model that prevent a perfect fitting of the deformed model to the query. This gap between the deformed model and the query induces mph{asymmetric} relationships among the models, which cannot be handled by typical metric learning techniques. Thus, to retrieve the best models for fitting, we propose a novel deep embedding approach that learns the asymmetric relationships by leveraging location-dependent egocentric distance fields. We also propose two strategies for training the embedding network. We demonstrate that both of these approaches outperform other baselines in our experiments with both synthetic and real data. Our project page can be found at \href{https://deformscan2cad.github.io/}{deformscan2cad.github.io}.