Multiscale Feature Learning Based on Enhanced Feature Pyramid for Vehicle Detection

Abstract

Vehicle detection is a crucial task in autonomous driving systems. Due to large variance of scales and heavy occlusion of vehicle in an image, this task is still a challenging problem. Recent vehicle detection methods typically exploit feature pyramid to detect vehicles at different scales. However, the drawbacks in the design prevent the multiscale features from being completely exploited. This paper introduces a feature pyramid architecture to address this problem. In the proposed architecture, an improving region proposal network is designed to generate intermediate feature maps which are then used to add more discriminative representations to feature maps generated by the backbone network, as well as improving the computational cost of the network. To generate more discriminative feature representations, this paper introduces multilayer enhancement module to reweight feature representations of feature maps generated by the backbone network to increase the discrimination of foreground objects and background regions in each feature map. In addition, an adaptive RoI pooling module is proposed to pool features from all pyramid levels for each proposal and fuse them for the detection network. Experimental results on the KITTI vehicle detection benchmark and the PASCAL VOC 2007 car dataset show that the proposed approach obtains better detection performance compared with recent methods on vehicle detection.

Medical education and artificial intelligence: Question answering for medical questions based on intelligent interaction

“…In view of the above representations, an interaction unit outputs three feature combinations, as shown in Equation (9).…”

“…Hoanh Nguyen et al used a feature pyramid architecture and an adaptive RoI pooling module to detect vehicles of different scales to solve the large‐scale differences and severe occlusion in vehicle detection. The results indicate that it has better detection performance and lower computational cost 9 . Han et al proposed a CNN‐M2R network with multilayer fusion and multidimensional attention to improve vehicle detection efficiency in urban areas.…”

Deep Learning Techniques for Vehicle Detection and Classification from Images/Videos: A Survey

“… PN + FTN + Fusion + Concant has higher overall mAP than Faster RCNN, PN + FTN + Fusion. Nguyen [ 83 ] KITTI benchmark. PASCAL VOC 07.…”

Video Surveillance Vehicle Detection Method Incorporating Attention Mechanism and YOLOv5

“…Wang et al aimed for improving the VD and tracking of autonomous vehicles using 3D Light Detection and Ranging (LiDAR) accuracy, a clustering algorithm trained by support vector machine (SVM) algorithm combined with Kalman filter and global nearest neighbor (GNN) algorithm is proposed to employ tracking of vehicles and further improve the accuracy of VD results with the help of tracking results [4]. Nguyen address the problem of large scale differences of vehicles and severe vehicle occlusion in VD by using a feature The results show better detection performance and lower computational cost [5]. Han et al propose a CNN-M2R network with multilayer fusion and multidimensional attention to improve VD performance in urban areas, which uses a multidimensional attention network to highlight target convergence and a new difficulty-positive and negative sample balanced sampling strategy and a global balanced loss function to handle spatial imbalance and objective imbalance, the experimental results show a great improvement in detection performance compared to SSD, LRTDet, RFCN, and DFPN [6].…”