Comparison of Yolo, SSD, Faster RCNN for Real Time Tennis Ball Tracking for Action Decision Networks
Authors:
- 1 April 2019
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-4
Abstract
This paper describes a systemic approach that analyses tennis videos to estimate its trajectory when the ball is tossed by the player. This system will reconstruct the trajectory of the ball by combining various image processing techniques to interpret the video frames using Action Decision networks. The project estimates the ball location using multiple-view geometry and state estimation filtering. Image processing concepts like image segmentation, morphological image processing are employed. We will perform the project using three different algorithms namely YOLO, SSD and Faster RCNN. A comparison is done using the three different algorithms and the performance of the different algorithms will be determined for the detection of a tennis ball. Software has been developed to compare the algorithms and to find the algorithm that is more efficient and has less computational power.Keywords
This publication has 6 references indexed in Scilit:
- Action-Decision Networks for Visual Tracking with Deep Reinforcement Learning2017
- SSD: Single Shot MultiBox Detector2016
- Faster R-CNN: Towards Real-Time Object Detection with Region Proposal NetworksIEEE Transactions on Pattern Analysis and Machine Intelligence, 2016
- You Only Look Once: Unified, Real-Time Object Detection2016
- MUlti-Store Tracker (MUSTer): A cognitive psychology inspired approach to object tracking2015
- Robust Object Tracking with Online Multiple Instance LearningIEEE Transactions on Pattern Analysis and Machine Intelligence, 2010