Dr. Vijaykrishnan Narayanan, Thesis Supervisor Dr. Vijaykrishnan Narayanan, Thesis Honors Advisor Chitaranjan Das, Faculty Reader
Keywords:
Computer Vision Sliding Window Region Proposal ROI FPGA Hardware Acceleration
Abstract:
Visual Object Searching is a task performed within a wide range of applications from surveillance to medical imaging. Locating a given object within a scene is commonly done through two methods: an exhaustive or semi – exhaustive search followed by classification through the entire scene, or region proposal followed by classification. In practice, reasonable classification accuracies, especially in real time systems, have been achieved by incorporating the latter to avoid expensive searching of the entire scene. However, the system’s performance becomes dependent on the region proposal scheme’s ability to produce favorable ROIs. With the aid of hardware acceleration, removal of this dependency is obtainable through an exhaustive or semi-exhaustive search of the image. This work aims to investigate the computational performance benefits of using either method, provided the opportunity for hardware acceleration.