Bounding overwatch

About: Bounding overwatch is a research topic. Over the lifetime, 966 publications have been published within this topic receiving 15156 citations.

The Design of Objects Bounding Boxes Non-Maximum Suppression and Visualization Module Based on FPGA

Abstract: The objects detection results are marked with bounding boxes in image. There are lots of redundant bounding boxes in detection results. Generally, the non-maximum suppression (NMS) method is used to remove redundant bounding boxes in detection results to improve accuracy of detection. Opencv is usually used to complete visualization of objects bounding boxes. We compare time of NMS and bounding boxes visualization implemented in FPGA and software in server. The results in hardware performs better. Objects detection has high requirements for real-time performance. Parallel computing capabilities of hardware makes it better than software to implement object detection algorithm and related peripheral algorithms in FPGA. FPGA is a good solution for embedded applications of objects detection. In this paper, we build a FPGA based objects detection system on YOLO, which is a kind of object detection algorithm based on deep learning and we complete the design and verification of back-end processing module in the system, including NMS module to remove redundant objects bounding boxes and bounding boxes visualization module. We make some changes for the NMS algorithm implementation in hardware. Two modules work together to process 100x61 images including 98x20 object bounding boxes. The remaining bounding boxes what we want after removing redundant bounding boxes are used for display, which ultimately takes 680μs.

Dual-split trees

Abstract: We introduce the dual-split tree, a new tree-based acceleration structure for ray tracing. Each internal node of a dual-split tree uses two axis-aligned planes to either split the parent node into two child nodes or to mark the empty regions of the node. This allows child bounding boxes to overlap when desired. Thus, our dual-split tree is capable of representing space partitioning identical to any given bounding volume hierarchy. Our dual-split tree provides a significant reduction in the required acceleration structure storage by eliminating the redundant bounding planes that are commonplace in bounding volume hierarchies, providing better performance and storage savings than similar previous methods. As a result, we achieve improved rendering performance with dual-split trees, as compared to bounding volume hierarchies with a comparable level of optimization using identical or similar space partitioning.

Barycentric Bounds in Stochastic Programming: Theory and Application

Abstract: The design and analysis of efficient approximation schemes are of fundamental importance in stochastic programming research Bounding approximations are particularly popular for providing strict error bounds that can be made small by using partitioning techniques In this chapter we develop a powerful bounding method for linear multistage stochastic programs with a generalized nonconvex dependence on the random parameters Thereby, we establish bounds on the recourse functions as well as compact bounding sets for the optimal decisions We further demonstrate that our bounding methods facilitate the reliable solution of important real-life decision problems To this end, we solve a stochastic optimization model for the management of nonmaturing accounts and compare the bounds on maximum profit obtained with different partitioning strategies

FreeSOLO: Learning to Segment Objects without Annotations

Abstract: Instance segmentation is a fundamental vision task that aims to recognize and segment each object in an image. However, it requires costly annotations such as bounding boxes and segmentation masks for learning. In this work, we propose a fully unsupervised learning method that learns class-agnostic instance segmentation without any annotations. We present FreeSOLO, a self-supervised instance segmentation framework built on top of the simple instance segmentation method SOLO. Our method also presents a novel localization-aware pre-training framework, where objects can be discovered from complicated scenes in an unsupervised manner. FreeSOLO achieves 9.8% $AP_{50}$ on the challenging COCO dataset, which even outperforms several segmentation proposal methods that use manual annotations. For the first time, we demonstrate unsupervised class-agnostic instance segmen-tation successfully. FreeSOLO's box localization significantly outperforms state-of-the-art unsupervised object de-tection/discovery methods, with about 100% relative improvements in COCO AP. FreeSOLO further demonstrates superiority as a strong pre-training method, outperforming state-of-the-art self-supervised pre-training methods by $+9.8\%$ AP when fine-tuning instance segmentation with only 5% COCO masks. Code is available at: github.com/NVlabs/FreeSOLO

Spatial-Temporal Action Localization With Hierarchical Self-Attention

Abstract: This paper proposes a novel architecture for spatial-temporal action localization in videos. The new architecture first employs a two-stream 3D convolutional neural network (3D-CNN) to provide initial action detection. Next, a new Hierarchical Self-Attention Network (HiSAN), the core of this architecture, is developed to learn the spatial-temporal relationships of key actors. Spatial Gaussian priors (SGP) are also imbued to the bidirectional self-attention to enhance HiSAN in modelling the relationships of neighboring actors. Such a combination of 3D-CNN and SGP augmented HiSAN allows us to effectively extract both of the spatial context information and the long-term temporal dependency to improve action localization accuracy. Afterwards, a new fusion strategy is employed, which first re-scores the bounding boxes to settle the inconsistent detection scores caused by background clutter or occlusion, and then aggregates the motion and appearance information from the two-stream network with the motion saliency to alleviate the impact of camera movement. Finally, a tube association network based on the self-similarity of the actors’ appearance and spatial information across frames is addressed to efficaciously construct the action tubes. Simulations on four widespread datasets reveal the efficacy of the new approach.