Improved CenterNet for Accurate and Fast Fitting Object Detection

Abstract

Accurate and fast detection of typical fittings is the prerequisite of condition monitoring and fault diagnosis. At present, most successful fitting detectors are anchor-based, which are challenging to meet the requirements of edge deployment. In this paper, we propose a novel anchor-free method called HRM-CenterNet. Firstly, the lightweight MobileNetV3 is introduced into CenterNet to extract multi-scale features of different layers. In addition, the lightweight receptive field enhancement module is proposed for the deep layer features to further enhance the characterization power of global features and generate more accurate heatmaps. Finally, the high-resolution feature fusion network with iterative aggregation is designed to reduce the loss of spatial semantic information in subsampling and further improve the accuracy of small and occlusion objects. Experiments are carried out on the TFITS and PASCAL VOC datasets. The results show that the size of the network is more than 60% lower than that of CenterNet. Compared with other detectors, our method achieves comparable accuracy with all accurate models at a much faster speed and meets the performance requirements of real-time detection.

Detection of Green Asparagus Using Improved Mask R-CNN for Automatic Harvesting

“…Compared with the original Faster RCNN, the improved version enhanced the accuracy of object detection [ 52 , 53 ]. In terms of fitting target detection, the HRM-CenterNet, which combined the lightweight MobileNetV3 and CenterNet, had a smaller network dimension and faster speed than the original CenterNet [ 54 ]. In [ 55 ], 15 types of vegetables were utilized to measure the effect of the DNN-framework-based recognition system on each vegetable category; in addition, the performance of learning accuracy and loss for a vegetable recognition system based on Caffe and Chainer frameworks was evaluated, which showed that the results of Caffe were better than those of Chainer.…”