Utilizing Temporal Information in Deep Convolutional Network for Efficient Soccer Ball Detection and Tracking in a Video
-
This project work is focued on the detection of soccer ball in an image using a CNN model which has been used in the Robo Soccer tournament. Further the work has been extended to track the ball in a sequence of frames using Convolutional LSTM.
-
The dataset used for thie project was generated on our own by capturing videos at the robotics laboratory and further the videos were split into frames and annotated using YOLO and further manually annotated.