WACV 2023 Papers

3D MRI imaging is based on a number of imaging sequences such as T1, T2, T1ce, and Flair, and each of them is performed by a group of two-dimensional scans. In practical MRI, some scans are often missing while many medical applications require a full set of scans. An MRI imputation method is presented, which synthesizes such missing scans. Key components in this method are the index registration and the intensity registration. The index registration models anatomical differences between two different scans in the same imaging sequence, and the intensity registration reflects the image contrast differences between two different scans of the same index. Two registration fields are learned to be invariant, and accordingly, allow two estimates of a missing scan, one within corresponding imaging sequence and another along scan index; the two estimates are combined to yield the final synthesized scan. Experimental results highlight that the proposed method improves prevalent limitations existing in previous synthesis methods, blending both structural and contrast aspects and capturing subtle parts of the brain. Quantitative results also show the superiority in various data sets, transitions, and measures.

In few-shot open-set recognition (FSOSR) for hyperspectral images (HSI), one major challenge arises due to the simultaneous presence of spectrally fine-grained known classes and outliers. Prior research on generative FSOSR cannot handle such a situation due to their inability to approximate the open space prudently. To address this issue, we propose a method, Meta-learning-based Open-set Recognition via Generative Adversarial Network (MORGAN), that can learn a finer separation between the closed and the open spaces. MORGAN seeks to generate class-conditioned adversarial samples for both the closed and open spaces in the few-shot regime using two GANs by judiciously tuning noise variance while ensuring discriminability using a novel Anti-Overlap Latent (AOL) regularizer. Adversarial samples from low noise variance amplify known class data density, and we use samples from high noise variance to augment known-unknowns. A first-order episodic strategy is adapted to ensure stability in the GAN training. Finally, we introduce a combination of metric losses which push these augmented known-unknowns or outliers to disperse in the open space while condensing known class distributions. Extensive experiments on four benchmark HSI datasets indicate that MORGAN achieves state-of-the-art FSOSR performance consistently.

Multi-camera tracking systems are gaining popularity in applications that demand high-quality tracking results, such as frictionless checkout. In cluttered and crowded environments, monocular multi-object tracking (MOT) systems often fail due to occlusions. Multiple highly overlapped cameras are capable of recovering partial 3D information. When used properly, 3D data can significantly alleviate the occlusion issue. However, training a multi-camera tracker demands a large-scale multi-camera tracking dataset with diverse camera settings and backgrounds. These requirements make the collection of multi-camera tracking dataset challenging and expensive. The cost of creating such a dataset has limited the availability and scale of datasets in this domain. Instead, we appeal to an auto-annotation system to reduce the cost. The system uses overlapped and calibrated depth and RGB cameras to build a 3D tracker and automatically generates the 3D tracking results. We then manually check and correct the 3D tracking results to ensure the label quality, which is much cheaper than solely manual annotation. Next, the 3D tracking results are projected to each calibrated RGB camera view to create 2D tracking results. In this way, we collect and annotate a large-scale densely labeled multi-camera tracking dataset from five different environments. We have conducted extensive experiments using two real-time multi-camera trackers and a person re-identification (ReID) model under different settings. This dataset provides a reliable benchmark for multi-camera, multi-object tracking systems in cluttered and crowded environments. We expect this benchmark to encourage more research attempts in this domain. Also, our results demonstrate that adapting the trackers and ReID models on this dataset significantly improves their performance. Our dataset will be publicly released upon the acceptance of this work.

In this paper, we introduce ML-Decoder, a new attention-based classification head. ML-Decoder predicts the existence of class labels via queries, and enables better utilization of spatial data compared to global average pooling. By redesigning the decoder architecture, and using a novel group-decoding scheme, ML-Decoder is highly efficient, and can scale well to thousands of classes. Compared to using a larger backbone, ML-Decoder consistently provides a better speed-accuracy trade-off. ML-Decoder is also versatile - it can be used as a drop-in replacement for various classification heads, and generalize to unseen classes when operated with word queries. Novel query augmentations further improve its generalization ability. Using ML-Decoder, we achieve state-of-the-art results on several classification tasks: on MS-COCO multi-label, we reach 91.1% mAP; on NUS-WIDE zero-shot, we reach 31.1% ZSL mAP; and on ImageNet single-label, we reach with vanilla ResNet50 backbone a new top score of 80.7%, without extra data or distillation. Public code will be available.

Recent research about camouflaged object detection (COD) aims to segment highly concealed objects hidden in complex surroundings. The tiny, fuzzy camouflaged objects result in visually indistinguishable properties. However, current single-view COD detectors are sensitive to background distractors. Therefore, blurred boundaries and variable shapes of the camouflaged objects are challenging to be fully captured with a single-view detector. To overcome these obstacles, we propose a behavior-inspired framework, called Multi-view Feature Fusion Network (MFFN), which mimics the human behaviors of finding indistinct objects in images, i.e., observing from multiple angles, distances, perspectives. Specifically, the key idea behind it is to generate multiple ways of observation (multi-view) by data augmentation and apply them as inputs. MFFN captures critical boundary and semantic information by comparing and fusing extracted multi-view features. In addition, our MFFN exploits the dependence and interaction between views and channels. Specifically, our methods leverage the complementary information between different views through a two-stage attention module called Co-attention of Multi-view (CAMV). And we design a local-overall module called Channel Fusion Unit (CFU) to explore the channel-wise contextual clues of diverse feature maps in an iterative manner. The experiment results show that our method performs favorably against existing state-of-the-art methods via training with the same data. The code will be available at https://github.com/dwardzheng/MFFN_COD.

Occlusions have long been a hard nut to crack in optical flow estimation due to ambiguous pixels matching between abutting images. Current methods only take two consecutive images as input, which is challenging to capture temporal coherence and reason about occluded regions. In this paper, we propose a novel optical flow estimation framework, namely MFCFlow, which attempts to compensate for the information of occlusions by mining and transferring motion features between multiple frames. Specifically, we construct a Motion-guided Feature Compensation cell (MFC cell) to enhance the ambiguous motion features according to the correlation of previous features obtained by attention-based structure. Furthermore, a TopK attention strategy is developed and embedded into the MFC cell to improve the subsequent matching quality. Extensive experiments demonstrate that our MFCFlow achieves significant improvements in occluded regions and attains state-of-the-art performances on both Sintel and KITTI benchmarks among other multi-frame optical flow methods.

In this paper, we present the Multi-view Extended Videos with Identities (MEVID) dataset for large-scale, video person re-identification (ReID) in the wild. To our knowledge, MEVID represents the most-varied video person ReID dataset, spanning an extensive indoor and outdoor environment across nine unique dates in a 73-day window, various camera viewpoints, and entity clothing changes. Specifically, we label the identities of 158 unique people wearing 598 outfits taken from 8,092 tracklets, average length of about 590 frames, seen in 33 camera views from the very-large-scale MEVA person activities dataset. While other datasets have more unique identities, MEVID emphasizes a richer set of information about each individual, such as: 4 outfits/identity vs. 2 outfits/identity in CCVID, 33 viewpoints across 17 locations vs. 6 in 5 simulated locations for MTA, and 10 million frames vs. 3 million for LS-VID. Being based on the MEVA video dataset, we also inherit data that is intentionally demographically balanced to the continental United States. To accelerate the annotation process, we developed a semi-automatic annotation framework and GUI that combines state-of-the-art real-time models for object detection, pose estimation, person ReID, and multi-object tracking. We evaluate several state-of-the-art methods on MEVID challenge problems and comprehensively quantify their robustness in terms of changes of outfit, scale, and background location. Our quantitative analysis on the realistic, unique aspects of MEVID shows that there are significant remaining challenges in video person ReID and indicates important directions for future research.

We propose MASTAF, a Model-Agnostic Spatio-Temporal Attention Fusion network for few-shot video classification. MASTAF takes input from a general video spatial and temporal representation,e.g., using 2D CNN, 3D CNN, and Video Transformer. Then, to make the most of such representations, we use self- and cross-attention models to highlight the critical spatio-temporal region to increase the inter-class variations and decrease the intra-class variations. Last, MASTAF applies a lightweight fusion network and a nearest neighbor classifier to classify each query video. We demonstrate that MASTAF improves the state-of-the-art performance on three few-shot video classification benchmarks(UCF101, HMDB51, and Something-Something-V2), e.g., by up to 91.6%, 69.5%, and 60.7% for five-way one-shot video classification, respectively.

Recently, neural network for scene flow estimation show impressive results on automotive data such as the KITTI benchmark. However, despite of using sophisticated rigidity assumptions and parametrizations, such networks are typically limited to only two frame pairs which does not allow them to exploit temporal information. In our paper we address this shortcoming by proposing a novel multi-frame approach that considers an additional preceding stereo pair. To this end, we proceed in two steps: Firstly, building upon the recent RAFT-3D approach, we develop an improved two-frame baseline by incorporating an advanced stereo method. Secondly, and even more importantly, exploiting the specific modeling concepts of RAFT-3D, we propose a U-Net architecture that performs a fusion of forward and backward flow estimates and hence allows to integrate temporal information on demand. Experiments on the KITTI benchmark do not only show that the advantages of the improved baseline and the temporal fusion approach complement each other, they also demonstrate that the computed scene flow is highly accurate. More precisely, our approach ranks second overall and first for the even more challenging foreground objects, in total outperforming the original RAFT-3D method by more than 16%. Code is available at https://github.com/cv-stuttgart/M-FUSE.

Recent work has shown a strong theoretical connection between variational autoencoders (VAEs) and the rate distortion theory. Motivated by this, we consider the problem of lossy image compression from the perspective of generative modeling. Starting from ResNet VAEs, which are originally designed for data (image) distribution modeling, we redesign their latent variable model using a quantization-aware posterior and prior, enabling easy quantization and entropy coding for image compression. Along with improved neural network blocks, we present a powerful and efficient class of lossy image coders, outperforming previous methods on natural image (lossy) compression. Our model compresses images in a coarse-to-fine fashion and supports parallel encoding and decoding, leading to fast execution on GPUs.

Due to the lack of training labels and the difficulty of annotating, dealing with adverse driving conditions such as nighttime has posed a huge challenge to the perception system of autonomous vehicles. Therefore, adapting knowledge from a labelled daytime domain to an unlabelled nighttime domain has been widely researched. In addition to labelled daytime datasets, existing nighttime datasets usually provide nighttime images with corresponding daytime reference images captured at nearby locations for reference. The key challenge is to minimize the performance gap between the two domains. In this paper, we propose LoopDA for domain adaptive nighttime semantic segmentation. It consists of self-loops that result in reconstructing the input data using predicted semantic maps, by rendering them into the encoded features. In a warm-up training stage, the self-loops comprise of an inner-loop and an outer-loop, which are responsible for intra-domain refinement and inter-domain alignment, respectively. To reduce the impact of day-night pose shifts, in the later self-training stage, we propose a co-teaching pipeline that involves an offline pseudo-supervision signal and an online reference-guided signal 'DNA' (Day-Night Agreement), bringing substantial benefits to enhance nighttime segmentation. Our model outperforms prior methods on Dark Zurich and Nighttime Driving datasets for semantic segmentation. Code and pretrained models are available at https://github.com/fy-vision/LoopDA.

Livestream videos have become a significant part of online learning, where design, digital marketing, creative painting, and other skills are taught by experienced experts in the sessions, making them valuable materials. However, Livestream tutorial videos are usually hours long, recorded, and uploaded to the Internet directly after the live sessions, making it hard for other people to catch up quickly. An outline will be a beneficial solution, which requires the video to be temporally segmented according to topics. In this work, we introduced a large Livestream video dataset named MultiLive, and formulated the temporal segmentation of the long Livestream videos (TSLLV) task. We propose LiveSeg, an unsupervised Livestream video temporal Segmentation solution, which takes advantage of multimodal features from different domains. Our method achieved a 16.8% F1-score performance improvement compared with the state-of-the-art method.

In this paper, we introduce LineEX that extracts data from scientific line charts. We adapt existing vision transformers and pose detection methods and showcase significant performance gains over existing SOTA baselines. We also propose a new loss function and present its effectiveness against existing loss functions. In addition, we synthetically created the largest line chart dataset comprising 430K images. The code and the dataset will be placed in the public domain soon after the acceptance.

Photorealistic style transfer is the task of synthesizing a realistic-looking image when adapting the content from one image to appear in the style of another image. Modern models commonly embed a transformation that fuses features describing the content image and style image and then decodes the resulting feature into a stylized image. We introduce a general-purpose transformation that enables controlling the balance between how much content is preserved and the strength of the infused style. We offer the first experiments that demonstrate the performance of existing transformations across different style transfer models, and demonstrate how transformation performs better in its ability to simultaneously run fast, produce consistently reasonable results, and control the balance between content and style in different models. To support reproducing our method and models, we share the code at https://github.com/chiutaiyin/LS-FT.

Video denoising is a fundamental problem in numerous computer vision applications. State-of-the-art attention-based denoising methods typically yield good results, but require vast amounts of GPU memory and usually suffer from very long computation times. Especially in the field of restoring digitized high-resolution historic films, these techniques are not applicable in practice. To overcome these issues, we introduce a lightweight video denoising network that combines efficient axial-coronal-sagittal (ACS) convolutions with a novel shifted window attention formulation (ASwin), which is based on the memory-efficient aggregation of self- and cross-attention across video frames. We numerically validate the performance and efficiency of our approach on synthetic Gaussian noise. Moreover, we train our network as a general-purpose blind denoising model for real-world videos, using a realistic noise synthesis pipeline to generate clean-noisy video pairs. A user study and non- reference quality assessment prove that our method outperforms the state-of-the-art on real-world historic videos in terms of denoising performance and temporal consistency.

Video motion magnification provides information to understand the subtle changes present in objects for applications like industrial, healthcare, sports, etc. Most state-ofthe-art (SOTA) methods use hand-crafted bandpass filters, which require prior information for the motion magnification, produces ringing artifacts, and small magnification in dynamic scenarios etc. While others use deep-learning based techniques, but their output suffers from artificially induced motion, distortions, blurriness, etc. Further, SOTA methods are computationally complex, which makes them less suitable for real-time applications. To address these problems, we proposed deep learning based simple yet effective solution for motion magnification. The proposed method uses a feature sharing and appearance encoder for better motion magnification with less distortions, artifacts etc. Additionally, for reducing magnification of noise and other unwanted changes, proxy-model based training is proposed. A computationally lightweight model ( 0.12 M parameters) is proposed along with the base model. The performance of the proposed models is tested qualitatively and quantitatively, with the SOTA methods. Results demonstrate the effectiveness of the proposed lightweight and base model over the existing SOTA methods.

Inspired by recent advances in vision transformers for object detection, we propose Li3DeTr, an end-to-end LiDAR based 3D Detection Transformer for autonomous driving, that inputs LiDAR point clouds and regresses 3D bounding boxes. The LiDAR local and global features are encoded using sparse convolution and multi-scale deformable attention respectively. In the decoder head, firstly, in the novel Li3DeTr cross-attention block, we link the LiDAR global features to 3D predictions leveraging the sparse set of object queries learnt from the data. Secondly, the object query interactions are formulated using multi-head self-attention. Finally, the decoder layer is repeated Ldec number of times to refine the object queries. Inspired by DETR, we employ set-to-set loss to train the Li3DeTr network. Without bells and whistles, the Li3DeTr network achieves 61.3% mAP and 67.6% NDS surpassing the state-of-the-art methods with non-maximum suppression (NMS) on the nuScenes dataset and it also achieves competitive performance on the KITTI dataset. We also employ knowledge distillation (KD) using a teacher and student model that slightly improves the performance of our network.

Fashion attribute editing is a task that aims to convert the semantic attributes of a given fashion image while preserving the irrelevant regions. Previous works typically employ conditional GANs where the generator explicitly learns the target attributes and directly execute the conversion. These approaches, however, are neither scalable nor generic as they operate only with few limited attributes and a separate generator is required for each dataset or attribute set. Inspired by the recent advancement of diffusion models, we explore the classifier-guided diffusion that leverages the off-the-shelf diffusion model pretrained on general visual semantics such as Imagenet. In order to achieve a generic editing pipeline, we pose this as multi-attribute image manipulation task, where the attribute ranges from item category, fabric, pattern to collar and neckline. We empirically show that conventional methods fail in our challenging setting, and study efficient adaptation scheme that involves recently introduced attention-pooling technique to obtain a multi-attribute classifier guidance. Based on this, we present a mask-free fashion attribute editing framework that leverages the classifier logits and the cross-attention map for manipulation. We empirically demonstrate that our framework achieves convincing sample quality and attribute alignments.

State-of-the-art generative model-based attacks against image classifiers overwhelmingly focus on single-object (ie., single dominant object) images. Different from such settings, we tackle a more practical problem of generating adversarial perturbations using multi-object (ie., multiple dominant objects) images as they are representative of most real-world scenes. Our goal is to design an attack strategy that can learn from such natural scenes by leveraging the local patch differences that occur inherently in such images (eg. difference between the local patch on the object 'person' and the object 'bike' in a traffic scene). Our key idea is to misclassify an adversarial multi-object image by confusing the victim classifier for each local patch in the image. Based on this, we propose a novel generative attack (called Local Patch Difference or LPD-Attack) where a novel contrastive loss function uses the aforesaid local differences in feature space of multi-object scenes to optimize the perturbation generator. Through various experiments across diverse victim convolutional neural networks, we show that our approach outperforms baseline generative attacks with highly transferable perturbations when evaluated under different white-box and black-box settings.