Gated CNN: Integrating multi-scale feature layers for object detection
作者:
Highlights:
• The “Gate” structure extracts powerful features for object detection.
• Two-branch structure predicts the locations and categories ofobjects respectively, where each branch learns different parameters for different tasks.
• The inter-class loss help detectors learn the discrepant information between categories and better differentiate similar objects of different categories
• The experimental results demonstrate that G-CNN outperforms the state-of-the-art approaches, with a mAP of 40.9% at 10.6 FPS.
摘要
•The “Gate” structure extracts powerful features for object detection.•Two-branch structure predicts the locations and categories ofobjects respectively, where each branch learns different parameters for different tasks.•The inter-class loss help detectors learn the discrepant information between categories and better differentiate similar objects of different categories•The experimental results demonstrate that G-CNN outperforms the state-of-the-art approaches, with a mAP of 40.9% at 10.6 FPS.
论文关键词:Gated CNN,object detection,multi-scale feature layers,explainable CNN
论文评审过程:Received 18 June 2019, Revised 24 October 2019, Accepted 24 November 2019, Available online 4 December 2019, Version of Record 5 June 2020.
论文官网地址:https://doi.org/10.1016/j.patcog.2019.107131