Convolutional neural network (CNN)-based stereo matching approaches generally require a dense cost volume (DCV) for disparity estimation. However, generating such cost volumes is computationally-intensive and memory-consuming, hindering CNN training and inference efficiency. To address this problem, we propose SCV-Stereo, a novel CNN architecture, capable of learning dense stereo matching from sparse cost volume (SCV) representations. Our inspiration is derived from the fact that DCV representations are somewhat redundant and can be replaced with SCV representations. Benefiting from these SCV representations, our SCV-Stereo can update disparity estimations in an iterative fashion for accurate and efficient stereo matching. Extensive experiments carried out on the KITTI Stereo benchmarks demonstrate that our SCV-Stereo can significantly minimize the trade-off between accuracy and efficiency for stereo matching. Our project page is https://sites.google.com/view/scv-stereo.
Stereo matching is a key component of autonomous driving perception. Recent unsupervised stereo matching approaches have received adequate attention due to their advantage of not requiring disparity ground truth. These approaches, however, perform poorly near occlusions. To overcome this drawback, in this paper, we propose CoT-Stereo, a novel unsupervised stereo matching approach. Specifically, we adopt a co-teaching framework where two networks interactively teach each other about the occlusions in an unsupervised fashion, which greatly improves the robustness of unsupervised stereo matching. Extensive experiments on the KITTI Stereo benchmarks demonstrate the superior performance of CoT-Stereo over all other state-of-the-art unsupervised stereo matching approaches in terms of both accuracy and speed. Our project webpage is https://sites.google.com/view/cot-stereo.
Regularization and data augmentation methods have been widely used and become increasingly indispensable in deep learning training. Researchers who devote themselves to this have considered various possibilities. But so far, there has been little discussion about regularizing outputs of the model. This paper begins with empirical observations that better performances are significantly associated with output distributions, that have smaller average values and variances. By audaciously assuming there is causality involved, we propose a novel regularization term, called Output Decay, that enforces the model to assign smaller and similar output values on each class. Though being counter-intuitive, such a small modification result in a remarkable improvement on performance. Extensive experiments demonstrate the wide applicability, versatility, and compatibility of Output Decay.
Self-supervised learning, which benefits from automatically constructing labels through pre-designed pretext task, has recently been applied for strengthen supervised learning. Since previous self-supervised pretext tasks are based on input, they may incur huge additional training overhead. In this paper we find that features in CNNs can be also used for self-supervision. Thus we creatively design the \emph{feature-based pretext task} which requires only a small amount of additional training overhead. In our task we discard different particular regions of features, and then train the model to distinguish these different features. In order to fully apply our feature-based pretext task in supervised learning, we also propose a novel learning framework containing multi-classifiers for further improvement. Original labels will be expanded to joint labels via self-supervision of feature transformations. With more semantic information provided by our self-supervised tasks, this approach can train CNNs more effectively. Extensive experiments on various supervised learning tasks demonstrate the accuracy improvement and wide applicability of our method.
Traditional self-supervised learning requires CNNs using external pretext tasks (i.e., image- or video-based tasks) to encode high-level semantic visual representations. In this paper, we show that feature transformations within CNNs can also be regarded as supervisory signals to construct the self-supervised task, called \emph{internal pretext task}. And such a task can be applied for the enhancement of supervised learning. Specifically, we first transform the internal feature maps by discarding different channels, and then define an additional internal pretext task to identify the discarded channels. CNNs are trained to predict the joint labels generated by the combination of self-supervised labels and original labels. By doing so, we let CNNs know which channels are missing while classifying in the hope to mine richer feature information. Extensive experiments show that our approach is effective on various models and datasets. And it's worth noting that we only incur negligible computational overhead. Furthermore, our approach can also be compatible with other methods to get better results.
The promising performances of CNNs often overshadow the need to examine whether they are doing in the way we are actually interested. We show through experiments that even over-parameterized models would still solve a dataset by recklessly leveraging spurious correlations, or so-called 'shortcuts'. To combat with this unintended propensity, we borrow the idea of printer test page and propose a novel approach called White Paper Assistance. Our proposed method involves the white paper to detect the extent to which the model has preference for certain characterized patterns and alleviates it by forcing the model to make a random guess on the white paper. We show the consistent accuracy improvements that are manifest in various architectures, datasets and combinations with other techniques. Experiments have also demonstrated the versatility of our approach on fine-grained recognition, imbalanced classification and robustness to corruptions.
Multiparty Dialogue Machine Reading Comprehension (MRC) differs from traditional MRC as models must handle the complex dialogue discourse structure, previously unconsidered in traditional MRC. To fully exploit such discourse structure in multiparty dialogue, we present a discourse-aware dialogue graph neural network, DADgraph, which explicitly constructs the dialogue graph using discourse dependency links and discourse relations. To validate our model, we perform experiments on the Molweni corpus, a large-scale MRC dataset built over multiparty dialogue annotated with discourse structure. Experiments on Molweni show that our discourse-aware model achieves statistically significant improvements compared against strong neural network MRC baselines.
This paper investigates two typical image-type representations for event camera-based tracking: time surface (TS) and event map (EM). Based on the original TS-based tracker, we make use of these two representations' complementary strengths to develop an enhanced version. The proposed tracker consists of a general strategy to evaluate the optimization problem's degeneracy online and then switch proper representations. Both TS and EM are motion- and scene-dependent, and thus it is important to figure out their limitations in tracking. We develop six tracker variations and conduct a thorough comparison of them on sequences covering various scenarios and motion complexities. We release our implementations and detailed results to benefit the research community on event cameras: https: //github.com/gogojjh/ESVO_extension.
With the recent advancement of deep learning technology, data-driven approaches for autonomous car prediction and planning have achieved extraordinary performance. Nevertheless, most of these approaches follow a non-interactive prediction and planning paradigm, hypothesizing that a vehicle's behaviors do not affect others. The approaches based on such a non-interactive philosophy typically perform acceptably in sparse traffic scenarios but can easily fail in dense traffic scenarios. Therefore, we propose an end-to-end interactive neural motion planner (INMP) for autonomous driving in this paper. Given a set of past surrounding-view images and a high definition map, our INMP first generates a feature map in bird's-eye-view space, which is then processed to detect other agents and perform interactive prediction and planning jointly. Also, we adopt an optical flow distillation paradigm, which can effectively improve the network performance while still maintaining its real-time inference speed. Extensive experiments on the nuScenes dataset and in the closed-loop Carla simulation environment demonstrate the effectiveness and efficiency of our INMP for the detection, prediction, and planning tasks. Our project page is at sites.google.com/view/inmp-ofd.
Recently, deep-learning based approaches have achieved impressive performance for autonomous driving. However, end-to-end vision-based methods typically have limited interpretability, making the behaviors of the deep networks difficult to explain. Hence, their potential applications could be limited in practice. To address this problem, we propose an interpretable end-to-end vision-based motion planning approach for autonomous driving, referred to as IVMP. Given a set of past surrounding-view images, our IVMP first predicts future egocentric semantic maps in bird's-eye-view space, which are then employed to plan trajectories for self-driving vehicles. The predicted future semantic maps not only provide useful interpretable information, but also allow our motion planning module to handle objects with low probability, thus improving the safety of autonomous driving. Moreover, we also develop an optical flow distillation paradigm, which can effectively enhance the network while still maintaining its real-time performance. Extensive experiments on the nuScenes dataset and closed-loop simulation show that our IVMP significantly outperforms the state-of-the-art approaches in imitating human drivers with a much higher success rate. Our project page is available at https://sites.google.com/view/ivmp.