The present invention discloses a method for assessing the activity of Takifugu
sperm cryopreservation solution based on
computer vision. [Solution] Takifugu
sperm are cryopreserved for long periods using a
cryopreservation solution, thawed, and then activated. After observation under a
microscope,
sperm motility videos are recorded. Data augmentation is performed on the video using a binarization method. Based on two
deep learning algorithms for target detection and target tracking, the positional changes of multiple Takifugu sperm targets between consecutive frames are tracked and captured, and the significant movements and trajectories of the recognized sperm are analyzed for
imitation training. The sperm's movement and viability are determined by statistically analyzing their displacement between different frames. Finally, a visualized sperm viability graph is generated. This method enables long-term
cryopreservation of highly active Takifugu sperm, providing
technical support for year-round
reproduction and species
resource conservation. It also improves the accuracy of Takifugu sperm
activity detection, simplifies the detection and recognition process, and reduces costs.