Search

Research on Object Detection Algorithm Based on Semantic Features

Author(s)

Shuili Zhang1,2, *, Xue Tian1, Kexin Zheng1

Affiliation(s)

1College of Physics and Electronic Information, Yan’ an University, Yan’An, Shaanxi, China 2Shaanxi Key Laboratory of Intelligent Processing for Big Energy Data, Yan’An, Shaanxi, China *Corresponding Author.

Abstract

In order to tackle the problem of low recognition accuracy for small objects or complex backgrounds due to the single and low-resolution feature maps extracted by the traditional Faster R-CNN network, this paper designs a target detection algorithm that combines semantic features with traditional features by integrating a U-Net semantic segmentation network with the Faster R-CNN network. To create region suggestions, the extracted convolutional and semantic features are first processed using the RPN network, these proposals are then unified to the same size. After fusing the features, the candidate boxes are brought back to identify the target items. This method addresses the issue of target object detection in complicated backdrops and the loss of small target features in deep feature maps. Lastly, the network model's resilience and capacity for generalization are increased by applying the soft-NMS method to improve the detection of overlapping regions. According to experimental results, this approach performs better overall and achieves higher detection accuracy on the COCO dataset when compared to the standard Faster R-CNN network.

Keywords

Faster R-CNN; Object Detection; Semantic Segmentation; U-Net

References

[1] Syed Sahil Abbas Zaidi, Mohammad Samar Ansari, Asra Aslam, Nadia Kanwal, Mamoona Asghar, et al. A survey of modern deep learning based object detection models. Digital Signal Processing, 2022, 126: 103514. [2] Shayhan Ameen Chowdhury, Mir Md. Saki Kowsar, Kaushik Deb. Human detection utilizing adaptive background mixture models and improved histogram of oriented gradients. ICT Express, 2018, 4 (4):216-220. [3] Girshick Ross, Donahue Jeff, Darrell Trevor, Malik Jitendra. Rich feature hierarchies for accurate object detection and semantic segmentation//Proceedings of the IEEE conference on computer vision and pattern recognition, 2014: 580-587. [4] Girshick, R..Fast R-CNN (Conference Paper).Proceedings of the IEEE International Conference on Computer Vision, 2015,Vol.2015: 1440-1448. [5] Ren Shaoqing, He Kaiming, Girshick Ross, Sun Jian. Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6):1137-1149. [6] Linxiu Wu, Houjie Li, Jianjun He, Xuan Chen. Traffic sign detection method based on Faster R-CNN//Journal of Physics: Conference Series, 2019, 1176(3): 032045. [7] Peiyuan Jiang, Daji Ergu, Fangyao Liu, Ying Cai, Bo Ma. A Review of Yolo Algorithm Developments. Procedia Computer Science, 2022, 199: 1066-1073. [8] Chenxi Zhang, Feng Kang, Yaxiong Wang. An Improved Apple Object Detection Method Based on Lightweight YOLOv4 in Complex Backgrounds. Remote Sensing, 2022, 14(17): 4150. [9] Hakan Bilen, Andrea Vedaldi. Weakly supervised deep detection networks//Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 2846-2854. [10] LiYang, Junhong Zhong, Yun Zhang, Sichang Bai, Guoshu Li, et al. An Improving Faster-RCNN with Multi-Attention ResNet for Small Target Detection in Intelligent Autonomous Transport with 6G. IEEE Transactions on Intelligent Transportation Systems. 2023, 24(7):P7717-7725. [11] Jiao Lin, Dong Shifeng, Zhang Shengyu, Xie Chengjun, Wang Hongqiang. AF-RCNN: An anchor-free convolutional neural network for multi-categories agricultural pest detection. Computers and Electronics in Agriculture, 2020, 174: 105522. [12] Hasib Zunair, A. Ben Hamza. Sharp U-Net: Depthwise convolutional network for biomedical image segmentation. Computers in Biology and Medicine, 2021, 136: 104699. [13] Xu Zhaozhuo, Xu Xin, Wang Lei, Yang Rui, Pu Fangling. Deformable convnet with aspect ratio constrained nms for object detection in remote sensing imagery. Remote Sensing, 2017, 9(12): 1312. [14] Navaneeth Bodla, Bharat Singh, Rama Chellappa, Larry S. Davis. Soft-NMS--improving object detection with one line of code//Proceedings of the IEEE international conference on computer vision. 2017: 5562-5570. [15] Prasanna Sheetal, ElSharkawy Mohamed. Improving Mean Average Precision (mAP) of Camera and Radar Fusion Network for Object Detection Using Radar Augmentation. Lecture Notes in Networks and Systems, 2023, 465:51-60.