Laser & Optoelectronics Progress, Volume. 56, Issue 4, 041003(2019)
Context-Sensitive Multi-Scale Face Detection
Figures & Tables(12)
Fig. 2. Context extraction module. (a)Stacked model; (b) context-sensitive module
Fig. 4. Precision-recall curves on Wider Face validation set. (a) Easy; (b) Medium; (c) Hard
Table 1. Comparison of context extraction module
View tableTable 1. Comparison of context extraction module
Model Stacked model CSMS context- sensitive module Number of parameters 5*3×3 3*3×3 Equivalent layer 3×3,5×5, 7×7 3*3×3,3*5×5, 7×7
Table 2. Comparison of mAP for scale invariance design%
View tableTable 2. Comparison of mAP for scale invariance design%
Level CSMS-sglDm CSMS Easy 81.2 92.6 Medium 83.6 91.0 Hard 39.8 82.5
Table 3. Comparison of mAP for context-sensitive module%
View tableTable 3. Comparison of mAP for context-sensitive module%
Level Stacked model CSMS Easy 92.5 92.6 Medium 90.8 91.0 Hard 82.0 82.5
Table 4. Comparison of mAP for training methods focused on hard sample samples%
View tableTable 4. Comparison of mAP for training methods focused on hard sample samples%
Level OHEM Focal Loss Easy 91.9 92.3 Medium 90.7 90.8 Hard 81.4 81.6
Table 5. Comparison of mAP for feature fusion%
View tableTable 5. Comparison of mAP for feature fusion%
Level CSMS-pl CSMS-plDlt CSMS Easy 92.3 92.6 92.6 Medium 90.8 90.8 91.0 Hard 82.0 82.2 82.5
Table 6. Ablation study of CSMS on Wider Face test set
View tableTable 6. Ablation study of CSMS on Wider Face test set
Method Contribution mAP / % Focal Loss Context-sensitive module Effective feature fusion Easy Medium Hard Baseline 91.9 90.7 81.4 CSMS √ √ √ 92.3 90.8 81.6 √ √ 92.4 90.8 82.2 √ 92.6 91.0 82.5
Table 7. Comparison of detection performance
View tableTable 7. Comparison of detection performance
Method mAP /% Easy Medium Hard Method in Ref. [15] 86.2 84.4 74.9 Method in Ref. [15] 92.5 91.0 80.6 Method in Ref. [14] 89.9 87.4 62.9 Method in Ref. [17] 91.9 90.7 81.4 Method in Ref. [17] 93.1 92.1 84.5 CSMS (VGG-16) 92.6 91.0 82.5 CSMS(VGG-16)+ Pyramid 93.8 92.1 85.1
Tools
Get Citation
Copy Citation Text
Long Chen, Yanwei Pang. Context-Sensitive Multi-Scale Face Detection[J]. Laser & Optoelectronics Progress, 2019, 56(4): 041003
Paper Information
Category: Image Processing
Received: Aug. 8, 2018
Accepted: Sep. 5, 2018
Published Online: Jul. 31, 2019
The Author Email: Chen Long (longchen@tju.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20