Feature-Selection-Based Attentional-Deconvolution Detector for German Traffic Sign Detection Benchmark

In this study, we propose a novel traffic sign detection algorithm based on the deeplearning approach. The proposed algorithm, which termed feature-selection-based attentionaldeconvolution detector (FSADD), is used along with “you look only once” (YOLO) v5 structure for feature selection. When applying selection inside network divides extracted maps after convolution layer into similar and non maps. Generally, obtained layers are outputs of filters random weights. Owing to randomness filter, obtains various kinds unnecessary components, degrades performance. However, grouping high similarities can increase relativeness each map, thereby improving specific targets from images. Furthermore, FSADD model has modified sizes receptive fields improved Many available general algorithms unsuitable German benchmark (GTSDB) because small these signs in Experimental comparisons were performed respect GTSDB show that comparable state-of-the-art while detecting 29 73.9% accuracy classification performances.