Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Novel Wide-Area Multiobject Detection System with High-Probability Region Searching

Xianlei Long, Hui Zhao, Chao Chen, Fuqiang Gu, Qingyi Gu

TL;DR

The paper tackles the challenge of reliable multiobject detection over large surveillance areas by marrying a wide-angle panoramic camera with a high-speed search camera, mediated by a panoramic probability map (PPM) and an uncertainty-aware probability searching module. By generating region-prior maps from panoramic segmentation and refining detections through adaptive particle sampling and probabilistic voting, the approach achieves high search efficiency and accurate localization. Experimental results in indoor and outdoor scenarios show improved recall and localization speed, with up to 120 fps processing and superior performance against baseline region-sampling methods. The contributions include a novel PPM-based region sampling method and a dynamic, uncertainty-driven search strategy that enable robust wide-area multiobject detection in challenging scenes.

Abstract

In recent years, wide-area visual surveillance systems have been widely applied in various industrial and transportation scenarios. These systems, however, face significant challenges when implementing multi-object detection due to conflicts arising from the need for high-resolution imaging, efficient object searching, and accurate localization. To address these challenges, this paper presents a hybrid system that incorporates a wide-angle camera, a high-speed search camera, and a galvano-mirror. In this system, the wide-angle camera offers panoramic images as prior information, which helps the search camera capture detailed images of the targeted objects. This integrated approach enhances the overall efficiency and effectiveness of wide-area visual detection systems. Specifically, in this study, we introduce a wide-angle camera-based method to generate a panoramic probability map (PPM) for estimating high-probability regions of target object presence. Then, we propose a probability searching module that uses the PPM-generated prior information to dynamically adjust the sampling range and refine target coordinates based on uncertainty variance computed by the object detector. Finally, the integration of PPM and the probability searching module yields an efficient hybrid vision system capable of achieving 120 fps multi-object search and detection. Extensive experiments are conducted to verify the system's effectiveness and robustness.

A Novel Wide-Area Multiobject Detection System with High-Probability Region Searching

TL;DR

The paper tackles the challenge of reliable multiobject detection over large surveillance areas by marrying a wide-angle panoramic camera with a high-speed search camera, mediated by a panoramic probability map (PPM) and an uncertainty-aware probability searching module. By generating region-prior maps from panoramic segmentation and refining detections through adaptive particle sampling and probabilistic voting, the approach achieves high search efficiency and accurate localization. Experimental results in indoor and outdoor scenarios show improved recall and localization speed, with up to 120 fps processing and superior performance against baseline region-sampling methods. The contributions include a novel PPM-based region sampling method and a dynamic, uncertainty-driven search strategy that enable robust wide-area multiobject detection in challenging scenes.

Abstract

In recent years, wide-area visual surveillance systems have been widely applied in various industrial and transportation scenarios. These systems, however, face significant challenges when implementing multi-object detection due to conflicts arising from the need for high-resolution imaging, efficient object searching, and accurate localization. To address these challenges, this paper presents a hybrid system that incorporates a wide-angle camera, a high-speed search camera, and a galvano-mirror. In this system, the wide-angle camera offers panoramic images as prior information, which helps the search camera capture detailed images of the targeted objects. This integrated approach enhances the overall efficiency and effectiveness of wide-area visual detection systems. Specifically, in this study, we introduce a wide-angle camera-based method to generate a panoramic probability map (PPM) for estimating high-probability regions of target object presence. Then, we propose a probability searching module that uses the PPM-generated prior information to dynamically adjust the sampling range and refine target coordinates based on uncertainty variance computed by the object detector. Finally, the integration of PPM and the probability searching module yields an efficient hybrid vision system capable of achieving 120 fps multi-object search and detection. Extensive experiments are conducted to verify the system's effectiveness and robustness.
Paper Structure (15 sections, 12 equations, 8 figures, 2 tables)

This paper contains 15 sections, 12 equations, 8 figures, 2 tables.

Table of Contents

  1. INTRODUCTION
  2. Hybrid Wide-Area Multiobject Detection System with Probability Searching
  3. PPM Generation for Region Sampling
  4. Semantic Segmentation-based PPM Generation
  5. Particle Sampling in High-Probability Regions
  6. High-probability Region Scanning Based on PPM
  7. Probability Searching for Object's Coordinate Refinement
  8. Coordinate Refinement
  9. Probability Searching
  10. Experiments
  11. Experimental Configuration
  12. Object Detection Analysis of the Proposed System
  13. Recall Comparison of Different Searching Methods
  14. Ablation Study of PPM and Probability Searching
  15. CONCLUSIONS

Figures (8)

  • Figure 1: Illustration of the proposed hybrid wide-area multiobject detection system. (a) shows the generation process of PPM, which includes three steps: panoramic image $I_w$ capture, semantic segmentation, and mask probability mapping. (b) is the proposed probability searching method, it includes galvano-mirror searching and scanning.
  • Figure 2: Visulization of high-probability region sampling and galvano-mirror scanning using particle filter.
  • Figure 3: Coordinate refinement $m$ by the probability searching. (a) The red rectangle on the left is the object's boundary after adjustment. (b) The red circle on the right is the sampling range during probability searching.
  • Figure 4: Construction of the system, which includes a wide-angle camera, a high-speed search camera, and a pan-tilt galvano-mirror.
  • Figure 5: Indoor object detection setup for performance verification.
  • ...and 3 more figures