This project involved the implementation of one of the most popular deep learning algorithms, a Convolutional Neural Network (CNN), to distinguish between images of fresh and infected salmon. The model developed using Python code in a Jupyter Notebook reached 92.2% of accuracy and 91.5% of precision when predicting classes on the test dataset.
You can check the Jupyter Notebook here: Salmon Scan Notebook
Here's an example of a sample image, with its true label displayed:
Dataset source: Ahmed, Md Shoaib (2024), “SalmonScan: A Novel Image Dataset for Machine Learning and Deep Learning Analysis in Fish Disease Detection in Aquaculture”, Mendeley Data, V3, doi: 10.17632/x3fz2nfm4w.3