A Pointer Meter Detection Method based on Optimal SSD network
Lei Lin, Zhihao Lu · 2019
Pointer meter detection is very useful for intelligent reading of pointer meter. This paper presents an optimal SSD network for pointer meter detection. The contributions includes: 1) the LBP image is introduced into the input layer of SSD network to enhance the adaptability of features to the environment; 2) the basic network of SSD is modified to extract multi-convolution fusion features through increasing a Fusing layer, and replacing the maxpooling method with S3Pool method, to enhance feature distinguishing ability for pointer meter detection. Experiments show that the new method has high TPR (True Positive Rate) and low FPR (False Positive Rate) for pointer meter detection.