Delving Deep into Multiscale Pedestrian Detection via Single Scale Feature Maps
The standard pipeline in pedestrian detection is sliding a pedestrian model on an image feature pyramid to detect pedestrians of different scales. In this pipeline, feature pyramid construction is time consuming and becomes the bottleneck for fast detection. Recently, a method called multiresolution...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2018-04-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | http://www.mdpi.com/1424-8220/18/4/1063 |
| _version_ | 1828359272728100864 |
|---|---|
| author | Xinchuan Fu Rui Yu Weinan Zhang Jie Wu Shihai Shao |
| author_facet | Xinchuan Fu Rui Yu Weinan Zhang Jie Wu Shihai Shao |
| author_sort | Xinchuan Fu |
| collection | DOAJ |
| description | The standard pipeline in pedestrian detection is sliding a pedestrian model on an image feature pyramid to detect pedestrians of different scales. In this pipeline, feature pyramid construction is time consuming and becomes the bottleneck for fast detection. Recently, a method called multiresolution filtered channels (MRFC) was proposed which only used single scale feature maps to achieve fast detection. However, there are two shortcomings in MRFC which limit its accuracy. One is that the receptive field correspondence in different scales is weak. Another is that the features used are not scale invariance. In this paper, two solutions are proposed to tackle with the two shortcomings respectively. Specifically, scale-aware pooling is proposed to make a better receptive field correspondence, and soft decision tree is proposed to relive scale variance problem. When coupled with efficient sliding window classification strategy, our detector achieves fast detecting speed at the same time with state-of-the-art accuracy. |
| first_indexed | 2024-04-14T03:40:22Z |
| format | Article |
| id | doaj.art-3cc99eacaf814de6b8b1a72d86ce10dd |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-04-14T03:40:22Z |
| publishDate | 2018-04-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-3cc99eacaf814de6b8b1a72d86ce10dd2022-12-22T02:14:33ZengMDPI AGSensors1424-82202018-04-01184106310.3390/s18041063s18041063Delving Deep into Multiscale Pedestrian Detection via Single Scale Feature MapsXinchuan Fu0Rui Yu1Weinan Zhang2Jie Wu3Shihai Shao4National Key Laboratory of Science and Technology on Communications, University of Electronic Science and Technology of China, Chengdu 611731, ChinaDepartment of Computer Science, University College London, London WC1E 6BT, UKDepartment of Computer Science & Engineering, Shanghai Jiao Tong University, Shanghai 200240, ChinaDepartment of MOE Research Center for Software/Hardware Co-Design Engineering and Application, East China Normal University, Shanghai 200062, ChinaNational Key Laboratory of Science and Technology on Communications, University of Electronic Science and Technology of China, Chengdu 611731, ChinaThe standard pipeline in pedestrian detection is sliding a pedestrian model on an image feature pyramid to detect pedestrians of different scales. In this pipeline, feature pyramid construction is time consuming and becomes the bottleneck for fast detection. Recently, a method called multiresolution filtered channels (MRFC) was proposed which only used single scale feature maps to achieve fast detection. However, there are two shortcomings in MRFC which limit its accuracy. One is that the receptive field correspondence in different scales is weak. Another is that the features used are not scale invariance. In this paper, two solutions are proposed to tackle with the two shortcomings respectively. Specifically, scale-aware pooling is proposed to make a better receptive field correspondence, and soft decision tree is proposed to relive scale variance problem. When coupled with efficient sliding window classification strategy, our detector achieves fast detecting speed at the same time with state-of-the-art accuracy.http://www.mdpi.com/1424-8220/18/4/1063pedestrian detectionboosted decision treescale invariancereceptive field correspondencesoft decision tree |
| spellingShingle | Xinchuan Fu Rui Yu Weinan Zhang Jie Wu Shihai Shao Delving Deep into Multiscale Pedestrian Detection via Single Scale Feature Maps Sensors pedestrian detection boosted decision tree scale invariance receptive field correspondence soft decision tree |
| title | Delving Deep into Multiscale Pedestrian Detection via Single Scale Feature Maps |
| title_full | Delving Deep into Multiscale Pedestrian Detection via Single Scale Feature Maps |
| title_fullStr | Delving Deep into Multiscale Pedestrian Detection via Single Scale Feature Maps |
| title_full_unstemmed | Delving Deep into Multiscale Pedestrian Detection via Single Scale Feature Maps |
| title_short | Delving Deep into Multiscale Pedestrian Detection via Single Scale Feature Maps |
| title_sort | delving deep into multiscale pedestrian detection via single scale feature maps |
| topic | pedestrian detection boosted decision tree scale invariance receptive field correspondence soft decision tree |
| url | http://www.mdpi.com/1424-8220/18/4/1063 |
| work_keys_str_mv | AT xinchuanfu delvingdeepintomultiscalepedestriandetectionviasinglescalefeaturemaps AT ruiyu delvingdeepintomultiscalepedestriandetectionviasinglescalefeaturemaps AT weinanzhang delvingdeepintomultiscalepedestriandetectionviasinglescalefeaturemaps AT jiewu delvingdeepintomultiscalepedestriandetectionviasinglescalefeaturemaps AT shihaishao delvingdeepintomultiscalepedestriandetectionviasinglescalefeaturemaps |