Object-Detection Competition

This project is an Object-Detection competition between Faster R-CNN and YOLO using PyTorch, Numpy, and Pandas

In this study, Alex and I tested the Faster R-CNN and YOLO models for Object Detection—a central problem in computer vision that is concerned with not only discerning the class of an object, but also discerning where the objects are located in an image. These models are the two prominent neural network architectures for this task.

Our results can be found in our report (Object_Detection), which includes our methodology, hypothesis, (qualitative and quantitative) results, discussion, insights on the two models (their architectures, features, pros/cons, etc), and our comprehensive conclusion.

Our code can be found in object-detection.ipynb

Example Results

The primary objective of our experiment was to both qualitatively and quantitatively evaluate the detection performance of the two models based on their object and boundary box classifications, as well as their total inference times.

We got a lot of very interesting insights based on our results, all of which can be found in the report.

Our study underscores the significance of both qualitative and quantitative assessments when selecting object detection models for specific applications. Furthermore, the results highlight the importance of considering the intended purpose and training strategies of these models in understanding their performance variations, which may vary depending on the specific task at hand.