Two-Stage Pedestrian Detection Model Using a New Classification Head for Domain Generalization

被引:2
作者
Schulz, Daniel [1 ,2 ,3 ]
Perez, Claudio A. [1 ,2 ,3 ]
机构
[1] Univ Chile, Dept Elect Engn, Santiago 8370451, Chile
[2] Univ Chile, Adv Min Technol Ctr, Santiago 8370451, Chile
[3] Ctr Intervent Med Precis & Adv Cellular Therapy, IMPACT, Santiago 7620086, Chile
关键词
pedestrian detection; domain generalization; object detection; triplet loss; two-stage detection; RECOGNITION;
D O I
10.3390/s23239380
中图分类号
O65 [分析化学];
学科分类号
070302 ; 081704 ;
摘要
Pedestrian detection based on deep learning methods have reached great success in the past few years with several possible real-world applications including autonomous driving, robotic navigation, and video surveillance. In this work, a new neural network two-stage pedestrian detector with a new custom classification head, adding the triplet loss function to the standard bounding box regression and classification losses, is presented. This aims to improve the domain generalization capabilities of existing pedestrian detectors, by explicitly maximizing inter-class distance and minimizing intra-class distance. Triplet loss is applied to the features generated by the region proposal network, aimed at clustering together pedestrian samples in the features space. We used Faster R-CNN and Cascade R-CNN with the HRNet backbone pre-trained on ImageNet, changing the standard classification head for Faster R-CNN, and changing one of the three heads for Cascade R-CNN. The best results were obtained using a progressive training pipeline, starting from a dataset that is further away from the target domain, and progressively fine-tuning on datasets closer to the target domain. We obtained state-of-the-art results, MR-2 of 9.9, 11.0, and 36.2 for the reasonable, small, and heavy subsets on the CityPersons benchmark with outstanding performance on the heavy subset, the most difficult one.
引用
收藏
页数:23
相关论文
共 50 条
[21]   A Two-Stage Model Compression Framework for Object Detection in Autonomous Driving Scenarios [J].
He, Qiyi ;
Xu, Ao ;
Ye, Zhiwei ;
Zhou, Wen ;
Zhang, Yifan ;
Xi, Ruijie .
IEEE SENSORS JOURNAL, 2025, 25 (02) :3735-3749
[22]   A two-stage approach to saliency detection in images [J].
Wang, Zheshen ;
Li, Baoxin .
2008 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING, VOLS 1-12, 2008, :965-968
[23]   A New Two-Stage Object Detection Network without RoI-Pooling [J].
Yang, Chao ;
Chen, Weihai ;
Chen, Peter C. Y. ;
Amezquita, Kendrick S. ;
Wu, Xingming .
PROCEEDINGS OF THE 30TH CHINESE CONTROL AND DECISION CONFERENCE (2018 CCDC), 2018, :1680-1685
[24]   A Two-Stage Deep-Learning Model for Detection and Occlusion-Based Classification of Kashmiri Orchard Apples for Robotic Harvesting [J].
Rathore, Divya ;
Divyanth, L. G. ;
Reddy, Kaamala Lalith Sai ;
Chawla, Yogesh ;
Buragohain, Mridula ;
Soni, Peeyush ;
Machavaram, Rajendra ;
Hussain, Syed Zameer ;
Ray, Hena ;
Ghosh, Alokesh .
JOURNAL OF BIOSYSTEMS ENGINEERING, 2023, 48 (02) :242-256
[25]   Floating object detection using double-labelled domain generalization [J].
Chen, Renfei ;
Peng, Yong ;
Li, Zhongwen ;
Shang, Hua .
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE, 2024, 133
[26]   Fusion-based speech emotion classification using two-stage feature selection [J].
Xie, Jie ;
Zhu, Mingying ;
Hu, Kai .
SPEECH COMMUNICATION, 2023, 152
[27]   A two-stage learning framework for imbalanced semi-supervised domain generalization fault diagnosis under unknown operating conditions [J].
Jian, Chuanxia ;
Chen, Heen ;
Ao, Yinhui ;
Zhang, Xiaobo .
ADVANCED ENGINEERING INFORMATICS, 2024, 62
[28]   Two-Stage Training of Graph Neural Networks for Graph Classification [J].
Do, Manh Tuan ;
Park, Noseong ;
Shin, Kijung .
NEURAL PROCESSING LETTERS, 2023, 55 (03) :2799-2823
[29]   A Two-Stage Neural Network Model for Automatic Detection of Skiffs from Satellite Imagery [J].
Ducaru, Tudor ;
Stoica, Stefan ;
Holban, Vlad ;
Minhas, Fayyaz .
2024 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS, IJCNN 2024, 2024,
[30]   Two-Stage Training of Graph Neural Networks for Graph Classification [J].
Manh Tuan Do ;
Noseong Park ;
Kijung Shin .
Neural Processing Letters, 2023, 55 :2799-2823