PIDNet: An Efficient Network for Dynamic Pedestrian Intrusion Detection

  title={PIDNet: An Efficient Network for Dynamic Pedestrian Intrusion Detection},
  author={Jingchen Sun and Jiming Chen and Tao Chen and Jiayuan Fan and Shibo He},
  journal={Proceedings of the 28th ACM International Conference on Multimedia},
Vision-based dynamic pedestrian intrusion detection (PID), judging whether pedestrians intrude an area-of-interest (AoI) by a moving camera, is an important task in mobile surveillance. The dynamically changing AoIs and a number of pedestrians in video frames increase the difficulty and computational complexity of determining whether pedestrians intrude the AoI, which makes previous algorithms incapable of this task. In this paper, we propose a novel and efficient multi-task deep neural network… 

Figures and Tables from this paper

How many Observations are Enough? Knowledge Distillation for Trajectory Forecasting

A novel distillation strategy that allows a knowledge transfer from a teacher network to a student one, the latter fed with fewer observations, and it is shown that a properly defined teacher supervision allows a student network to perform comparably to state-of-the-art approaches that demand more observations.



Pedestrian intrusion detection based on improved GMM and SVM

This paper focuses on the research work of dynamic pedestrian intrusion detection, improving some shortcomings in traditional methods, and takes advantage of the least squares fitting to optimize the motion area, making the detection rate greatly improved.

Perimeter intrusion detection based on intelligent video analysis

Fourier Descriptor and Histogram of Oriented Gradients are introduced to realize an effective detection of human bodies with multiple postures captured by fixed cameras for perimeter intrusion detection.

Moving-Object Intrusion Detection Based on Retinex-Enhanced Method

Experimental results show that the proposed Retinex-enhanced approach can provide an intrusion-detection rate of over 97% at daytime, over 90% at nighttime, and over 98% in the raining situation.

Small-scale Pedestrian Detection Based on Somatic Topology Localization and Temporal Feature Aggregation

A novel method integrated with somatic topological line localization (TLL) and temporal feature aggregation for detecting multi-scale pedestrians and a post-processing scheme based on Markov Random Field (MRF) is introduced to eliminate ambiguities in occlusion cases.

SSA-CNN: Semantic Self-Attention CNN for Pedestrian Detection

This work proposes a method that explores semantic segmentation results as self-attention cues to significantly improve the pedestrian detection performance and incorporates semantic attention information from multi-scale layers into deep convolution neural network to boost pedestrian detection.

Automated Intrusion Detection for Video Surveillance Using Conditional Random Fields

This paper proposes a new training algorithm for CRFs based on expectation maximization, which can be used with unlabeled data and applies the resulting trained CRF to separate normal activities from suspicious behavior.

ThunderNet: Towards Real-Time Generic Object Detection on Mobile Devices

benefit from the highly efficient backbone and detection part design, ThunderNet surpasses previous lightweight one-stage detectors with only 40% of the computational cost on PASCAL VOC and COCO benchmarks.

DetNet: Design Backbone for Object Detection

DetNet is proposed, which is a novel backbone network specifically designed for object detection that includes the extra stages against traditional backbone network for image classification, while maintains high spatial resolution in deeper layers.

A mobile vision system for robust multi-person tracking

A mobile vision system for multi-person tracking in busy environments that integrates continuous visual odometry computation with tracking-by-detection in order to track pedestrians in spite of frequent occlusions and egomotion of the camera rig is presented.

Tiny-DSOD: Lightweight Object Detection for Resource-Restricted Usages

Tiny-DSOD is proposed, based on the deeply supervised object detection (DSOD) framework, which introduces two innovative and ultra-efficient architecture blocks: depthwise dense block (DDB) based backbone and depthwise feature-pyramid-network (D-FPN) based front-end.