Abstract
As a hot topic in artificial intelligence, dog face recognition has been widely studied and applied in computer vision. The single shot detector (SSD) is one of the commonly used algorithms in computer vision, which can detect multi-scale objects accurately and quickly. However, the accuracy of SSD is not good when it is used to detect small objects. In this paper, we propose to add and optimize both dilated convolution (DC) and attention mechanisms to improve the SSD network structure, and we call it DC-Attention-SSD. The experimental results show that the DC-Attention-SSD on the ImageNet dataset achieves an accuracy of detection up to 99.4%, which is significantly higher than SSD. This is especially the case for the recognition of small objects and occluded faces, where recognition of the object has been significantly improved.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
