Automatic Fish Detection and Tracking in Underwater Videos
09/2017 - 08/2018
Advisor: Dr. Jung-Hua Wang
· Re-trained the Faster-RCNN-Inception-V2-COCO model from Google's TensorFlow Object Detection API and employed OpenCV with Python to detect and track fish in complex, changeable underwater environments from recorded video clips
· Collected and labeled fish pictures manually to create a custom training dataset
· Calculated the FPS to evaluate the model performance
˙ https://yijing-sie.github.io/#project