Crown-starfish-detection

Dataset

The Dataset was acquired from 17 YouTube, at the end we got 725 images.And using data augmentation we increased the number of images to 2548.

Pipline

• Image annotation
• Image Preprocessing: we used white balance enhancement technique • lighten dark images • solve the greenish-blue appearance

• Image Augmentation

Models Results

-Basic Model

-Preprocessed Model

-Augmented Model

DEMO

we used Flak API to deploy our model.

steps to run the code:

• Clone the repo
• pip install requirements.txt
• Run the application file " ../Crown-starfish-detection-with-Yolo5/yolov5/app.py"
• click the link that will appear and choose an image or vedio
• The results will be saved in " ../Crown-starfish-detection-with-Yolo5/results"

Note :

" you will need to change the paths in " ..\Crown-starfish-detection-with-Yolo5\yolov5\app.py" and " ..\Crown-starfish-detection-with-Yolo5\yolov5\detect.py" to your absolute path