Figure 1: In object detection datasets, the ground-truth bounding boxes have inherent ambiguities in some cases. The bounding box regressor is expected to get smaller loss from ambiguous bounding boxes with our KL Loss. (a)(c) The ambiguities introduced by inaccurate labeling. (b) The ambiguities introduced by occlusion. (d) The object boundary itself is ambiguous. It is unclear where the left boundary of the train is because the tree partially occludes it. (better viewed in color) AbstractLarge-scale object detection datasets (e.g., MS-COCO) try to define the ground truth bounding boxes as clear as possible. However, we observe that ambiguities are still introduced when labeling the bounding boxes. In this paper, we propose a novel bounding box regression loss for learning bounding box transformation and localization variance together. Our loss greatly improves the localization accuracies of various architectures with nearly no additional computation. The learned localization variance allows us to merge neighboring bounding boxes during non-maximum suppression (NMS), which further improves the localization performance. On MS-COCO, we boost the Average Precision (AP) of VGG-16 Faster R-CNN from 23.6% to 29.1%. More importantly, for ResNet-50-FPN Mask R-CNN, our method improves the AP and AP 90 by 1.8% and 6.2% respectively, which significantly outperforms previous stateof-the-art bounding box refinement methods. Our code and models are available at github.com/yihui-he/KL-Loss arXiv:1809.08545v3 [cs.CV]

No abstract

3D multi-object tracking (MOT) is essential to applications such as autonomous driving. Recent work focuses on developing accurate systems giving less attention to computational cost and system complexity. In contrast, this work proposes a simple real-time 3D MOT system with strong performance. Our system first obtains 3D detections from a LiDAR point cloud. Then, a straightforward combination of a 3D Kalman filter and the Hungarian algorithm is used for state estimation and data association. Additionally, 3D MOT datasets such as KITTI evaluate MOT methods in 2D space and standardized 3D MOT evaluation tools are missing for a fair comparison of 3D MOT methods. We propose a new 3D MOT evaluation tool along with three new metrics to comprehensively evaluate 3D MOT methods. We show that, our proposed method achieves strong 3D MOT performance on KITTI and runs at a rate of 207.4 FPS on the KITTI dataset, achieving the fastest speed among modern 3D MOT systems. Our code is publicly available at http://www.xinshuoweng.com/projects/AB3DMOT.

The potential of zero charge (Epzc) of electrodes can greatly influence the salt removal capacity, charge efficiency and cyclic stability of capacitive deionization (CDI). Thus optimizing the Epzc of CDI electrodes is of great importance. A simple strategy to negatively shift the Epzc of CDI electrodes by modifying commercial activated carbon with quaternized poly (4-vinylpyridine) (AC-QPVP) is reported in this work. The Epzc of the prepared AC-QPVP composite electrode is as negative as -0.745 V vs. Ag/AgCl. Benefiting from the optimized Epzc of electrodes, the asymmetric CDI cell which consists of the AC-QPVP electrode and a nitric acid treated activated carbon (AC-HNO3) electrode exhibits excellent CDI performance. For inverted CDI, the working potential window of the asymmetric CDI cell can reach 1.4 V, and its salt removal capacity can be as high as 9.6 mg/g. For extended voltage CDI, the salt removal capacity of the asymmetric CDI cell at 1.2/-1.2 V is 20.6 mg/g, which is comparable to that of membrane CDI using pristine activated carbon as the electrodes (19.5 mg/g). The present work provides a simple method to prepare highly positively charged CDI electrodes and may pave the way for the development of high-performance CDI cells.

No abstract