Project Name: Experiment Randomized Prim’s Algorithm for Object Proposals
Reference Paper: https://www.vision.ee.ethz.ch/publications/papers/proceedings/eth_biwi_01061.pdf
Source Code: https://github.com/smanenfr/rp
Report: https://github.com/Li-Yun/Randomized_Prim_Algorithm_Experiment/blob/master/ML_project_report.pdf
Summary: This project is part of Prof. Mitchell’s research project, and it was to empirically investigate how effective the object proposal method to reach completed dog walking situations. The completed dog walking situation represents dog walkers, dogs, and leashes should be found by machine learning algorithms. The idea of object proposal is the algorithms generate multiple bounding boxes for any possible object for a given image.
Duties:
- Read a research paper about leveraging an object proposal algorithm (randomized Prim's algorithm) to accomplish the dog walking task for a given definition of completed dog walking.
- Wrote experiment scripts in MATLAB to run the object proposal algorithm to evaluate how accurate the randomized Prim's algorithm is in the dog walking tasks.
- Wrote a summary report for the research project.