Automatic detection of follicle ultrasound images based on improved Faster R-CNN

Abstract

Follicle Ultrasonic images detection technology plays an important role in the monitoring of bull-follicle. Because of follicle ultrasound image containing lots of speckle noise and fuzzy edges, the traditional image detection algorithm is difficult to get better detection results on the ultrasonic image, and the traditional image detection algorithm needs to carry out sample feature extraction for each image, which is time-consuming and time-consuming and labor-intensive. According to the characteristics of the cattle follicle ultrasound image sets, this paper proposes a model of image detection based on improved deep learning Faster R - CNN to automatically detect cattle ovarian follicles, through joint VGG-16 different network layer characteristic figure to replace single deepest characteristic figure, retain the deep semantic characteristics at the same time, also keep the shallow characterization information. The experimental results show that this method has a better effect on the ultrasonic image detection of bovine follicle.