结合混合注意力机制的客滚船危险品检测研究OA

To address the issue of traditional instrument equipment and manual inspection being the main reliance for danger-ous goods detection on Ro/Ro passenger ships,a Faster RCNN algorithm combining hybrid attention mechanism is proposed.First-ly,the deep residual network ResNet50 is introduced to replace the Faster RCNN network's VGG16 for feature extraction.Then,a hybrid attention mechanism is introduced after the region generation network,aiming to mine spatiotemporal information and im-prove detection and classification performance.A large number of experimental results show that compared to existing object detec-tion algorithms,the proposed algorithm has better classification performance for dangerous goods detection,with an average classifi-cation result of 90.27%.