Object_detection_using_Yolo

A project for the detection of objects using Yolo

This repository is the subsection of the Project proposed, Image-Adaptive YOLO for Object Detection in Adverse Weather Conditions

The repository consists of working code for the Object Detection using YOLO.

Technical Papers

Study Papers can be found at the below location:

1. Yolo Paper
2. Image Adaptive Yolo Paper

Steps of executing the Yolo Detection program

1. Converting the darknet weights to tensorflow: Execute the save_model.py This step will save the weights on the local machine in the "checkpoints" folder

2. To execute the object detection on Image files: Execute the detect.py We need to send the parameters so the python file can select the appropriate model Sample Execution statement: python detect.py --weights ./checkpoints/yolov4-416 --size 416 --model yolov4 --images ./data/images/catsAndDogs.jpg

3. To execute the object detection on Video files: Execute the detect_video.py We need to send the parameters so the python file can select the appropriate model Sample Execution statement: python detect_video.py --weights ./checkpoints/yolov4-416 --size 416 --model yolov4 --video ./data/video/Chicago.mp4 --output ./detections/outputOfChicago_demo.mp4

Learning Goal

The outcome of this project, gave me following insights:

• Successfully used "OpenCv" for the image processing
• Understood the Bounding Boxes and ROI (Region of Interest)
• Was able to understand the Convolutional layers

Further work

• Understanding the advanced Computer Vision for the Image pre-processing
• Enhanching the characteristics of Yolo to detect objects in low-light and foggy images.