When applying multi-instance learning (MIL) to make predictions for bags of instances, the prediction accuracy of an instance often depends on not only the instance itself but also its context in the corresponding bag. From the viewpoint of causal inference, such bag contextual prior works as a confounder and may result in model robustness and interpretability issues. Focusing on this problem, we propose a novel interventional multi-instance learning (IMIL) framework to achieve deconfounded instance-level prediction. Unlike traditional likelihood-based strategies, we design an Expectation-Maximization (EM) algorithm based on causal intervention, providing a robust instance selection in the training phase and suppressing the bias caused by the bag contextual prior. Experiments on pathological image analysis demonstrate that our IMIL method substantially reduces false positives and outperforms state-of-the-art MIL methods.
Much progress has been made in the deep neural network (DNN) based diagnosis of mass lesions breast ultrasound (BUS) images. However, the non-mass lesion is less investigated because of the limited data. Based on the insight that mass data is sufficient and shares the same knowledge structure with non-mass data of identifying the malignancy of a lesion based on the ultrasound image, we propose a novel transfer learning framework to enhance the generalizability of the DNN model for non-mass BUS with the help of mass BUS. Specifically, we train a shared DNN with combined non-mass and mass data. With the prior of different marginal distributions in input and output space, we employ two domain alignment strategies in the proposed transfer learning framework with the insight of capturing domain-specific distribution to address the issue of domain shift. Moreover, we propose a cross-domain semantic-preserve data generation module called CrossMix to recover the missing distribution between non-mass and mass data that is not presented in training data. Experimental results on an in-house dataset demonstrate that the DNN model trained with combined data by our framework achieves a 10% improvement in AUC on the malignancy prediction task of non-mass BUS compared to training directly on non-mass data.
Previous deep learning based Computer Aided Diagnosis (CAD) system treats multiple views of the same lesion as independent images. Since an ultrasound image only describes a partial 2D projection of a 3D lesion, such paradigm ignores the semantic relationship between different views of a lesion, which is inconsistent with the traditional diagnosis where sonographers analyze a lesion from at least two views. In this paper, we propose a multi-task framework that complements Benign/Malignant classification task with lesion recognition (LR) which helps leveraging relationship among multiple views of a single lesion to learn a complete representation of the lesion. To be specific, LR task employs contrastive learning to encourage representation that pulls multiple views of the same lesion and repels those of different lesions. The task therefore facilitates a representation that is not only invariant to the view change of the lesion, but also capturing fine-grained features to distinguish between different lesions. Experiments show that the proposed multi-task framework boosts the performance of Benign/Malignant classification as two sub-tasks complement each other and enhance the learned representation of ultrasound images.
Natural videos provide rich visual contents for self-supervised learning. Yet most existing approaches for learning spatio-temporal representations rely on manually trimmed videos, leading to limited diversity in visual patterns and limited performance gain. In this work, we aim to learn representations by leveraging more abundant information in untrimmed videos. To this end, we propose to learn a hierarchy of consistencies in videos, i.e., visual consistency and topical consistency, corresponding respectively to clip pairs that tend to be visually similar when separated by a short time span and share similar topics when separated by a long time span. Specifically, a hierarchical consistency learning framework HiCo is presented, where the visually consistent pairs are encouraged to have the same representation through contrastive learning, while the topically consistent pairs are coupled through a topical classifier that distinguishes whether they are topic related. Further, we impose a gradual sampling algorithm for proposed hierarchical consistency learning, and demonstrate its theoretical superiority. Empirically, we show that not only HiCo can generate stronger representations on untrimmed videos, it also improves the representation quality when applied to trimmed videos. This is in contrast to standard contrastive learning that fails to learn appropriate representations from untrimmed videos.
Channel pruning has been broadly recognized as an effective technique to reduce the computation and memory cost of deep convolutional neural networks. However, conventional pruning methods have limitations in that: they are restricted to pruning process only, and they require a fully pre-trained large model. Such limitations may lead to sub-optimal model quality as well as excessive memory and training cost. In this paper, we propose a novel Channel Exploration methodology, dubbed as CHEX, to rectify these problems. As opposed to pruning-only strategy, we propose to repeatedly prune and regrow the channels throughout the training process, which reduces the risk of pruning important channels prematurely. More exactly: From intra-layer's aspect, we tackle the channel pruning problem via a well known column subset selection (CSS) formulation. From inter-layer's aspect, our regrowing stages open a path for dynamically re-allocating the number of channels across all the layers under a global channel sparsity constraint. In addition, all the exploration process is done in a single training from scratch without the need of a pre-trained large model. Experimental results demonstrate that CHEX can effectively reduce the FLOPs of diverse CNN architectures on a variety of computer vision tasks, including image classification, object detection, instance segmentation, and 3D vision. For example, our compressed ResNet-50 model on ImageNet dataset achieves 76% top1 accuracy with only 25% FLOPs of the original ResNet-50 model, outperforming previous state-of-the-art channel pruning methods. The checkpoints and code are available at here .
We present a novel framework named PlaneMVS for 3D plane reconstruction from multiple input views with known camera poses. Most previous learning-based plane reconstruction methods reconstruct 3D planes from single images, which highly rely on single-view regression and suffer from depth scale ambiguity. In contrast, we reconstruct 3D planes with a multi-view-stereo (MVS) pipeline that takes advantage of multi-view geometry. We decouple plane reconstruction into a semantic plane detection branch and a plane MVS branch. The semantic plane detection branch is based on a single-view plane detection framework but with differences. The plane MVS branch adopts a set of slanted plane hypotheses to replace conventional depth hypotheses to perform plane sweeping strategy and finally learns pixel-level plane parameters and its planar depth map. We present how the two branches are learned in a balanced way, and propose a soft-pooling loss to associate the outputs of the two branches and make them benefit from each other. Extensive experiments on various indoor datasets show that PlaneMVS significantly outperforms state-of-the-art (SOTA) single-view plane reconstruction methods on both plane detection and 3D geometry metrics. Our method even outperforms a set of SOTA learning-based MVS methods thanks to the learned plane priors. To the best of our knowledge, this is the first work on 3D plane reconstruction within an end-to-end MVS framework.
Vision-language representation learning largely benefits from image-text alignment through contrastive losses (e.g., InfoNCE loss). The success of this alignment strategy is attributed to its capability in maximizing the mutual information (MI) between an image and its matched text. However, simply performing cross-modal alignment (CMA) ignores data potential within each modality, which may result in degraded representations. For instance, although CMA-based models are able to map image-text pairs close together in the embedding space, they fail to ensure that similar inputs from the same modality stay close by. This problem can get even worse when the pre-training data is noisy. In this paper, we propose triple contrastive learning (TCL) for vision-language pre-training by leveraging both cross-modal and intra-modal self-supervision. Besides CMA, TCL introduces an intra-modal contrastive objective to provide complementary benefits in representation learning. To take advantage of localized and structural information from image and text input, TCL further maximizes the average MI between local regions of image/text and their global summary. To the best of our knowledge, ours is the first work that takes into account local structure information for multi-modality representation learning. Experimental evaluations show that our approach is competitive and achieves the new state of the art on various common down-stream vision-language tasks such as image-text retrieval and visual question answering.
Aligning signals from different modalities is an important step in vision-language representation learning as it affects the performance of later stages such as cross-modality fusion. Since image and text typically reside in different regions of the feature space, directly aligning them at instance level is challenging especially when features are still evolving during training. In this paper, we propose to align at a higher and more stable level using cluster representation. Specifically, we treat image and text as two "views" of the same entity, and encode them into a joint vision-language coding space spanned by a dictionary of cluster centers (codebook). We contrast positive and negative samples via their cluster assignments while simultaneously optimizing the cluster centers. To further smooth out the learning process, we adopt a teacher-student distillation paradigm, where the momentum teacher of one view guides the student learning of the other. We evaluated our approach on common vision language benchmarks and obtain new SoTA on zero-shot cross modality retrieval while being competitive on various other transfer tasks.
Pedestrian trajectory prediction is an essential component in a wide range of AI applications such as autonomous driving and robotics. Existing methods usually assume the training and testing motions follow the same pattern while ignoring the potential distribution differences (e.g., shopping mall and street). This issue results in inevitable performance decrease. To address this issue, we propose a novel Transferable Graph Neural Network (T-GNN) framework, which jointly conducts trajectory prediction as well as domain alignment in a unified framework. Specifically, a domain-invariant GNN is proposed to explore the structural motion knowledge where the domain-specific knowledge is reduced. Moreover, an attention-based adaptive knowledge learning module is further proposed to explore fine-grained individual-level feature representations for knowledge transfer. By this way, disparities across different trajectory domains will be better alleviated. More challenging while practical trajectory prediction experiments are designed, and the experimental results verify the superior performance of our proposed model. To the best of our knowledge, our work is the pioneer which fills the gap in benchmarks and techniques for practical pedestrian trajectory prediction across different domains.
Video instance segmentation (VIS) task requires classifying, segmenting, and tracking object instances over all frames in a video clip. Recently, VisTR has been proposed as end-to-end transformer-based VIS framework, while demonstrating state-of-the-art performance. However, VisTR is slow to converge during training, requiring around 1000 GPU hours due to the high computational cost of its transformer attention module. To improve the training efficiency, we propose Deformable VisTR, leveraging spatio-temporal deformable attention module that only attends to a small fixed set of key spatio-temporal sampling points around a reference point. This enables Deformable VisTR to achieve linear computation in the size of spatio-temporal feature maps. Moreover, it can achieve on par performance as the original VisTR with 10$\times$ less GPU training hours. We validate the effectiveness of our method on the Youtube-VIS benchmark. Code is available at https://github.com/skrya/DefVIS.