Sperm tracking and automatic extraction method based on machine learning

• Main Authors: LIU RAN, YAN YIMO, YUAN GUOHUI, HE JINHUI, WANG ZHUORAN, ZHANG BOXUAN, XIAO JIAN
• Format: Patent
• Language: Chinese; English
• Published: 25.03.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

The invention belongs to the technical field of sperm monitoring and automatic extraction, and provides a sperm tracking and automatic extraction method based on machine learning. The objective of the invention is to solve the problems of low monitoring precision, slow monitoring speed and unfavorable sperm multi-target real-time tracking existing in the current main sperm monitoring method. The main scheme comprises the following steps: acquiring real-time image data of a sperm sample; multi-sperm tracking is realized through a YOLO-V4, a KCF (Kalman filter) tracker and a Kalman filter; calculating the speed of the sperms according to the tracking result, and determining target sperms according to the speed; detecting and extracting the position of the needle in the image by using YOLO-V4; constructing an observation state of an extraction needle controller according to the position of the extraction needle and the tracking position of the sperm; and inputting the observation state into an actor network in a