Fruit Recognizer

Goal

Recognize fruits! Because, why not?

Dataset

It's currently using the Fruits-360 dataset

Challenges

The models I've been trying are doing great validation and training accuracy wise, the fruit 360 dataset is very limited only containing 7 fruits in total, therefore I couldn't test it with fruits that look alike, however it's having a really hard time distinguishing a zucchini from a cucumber (I do too).

Directory structure

• /models - All trained models are saved here with some information about them on the title.
• main.py - Holds the current code for training the model
• predicter.py - Run this to predict results on a given image or generator
• Figures_{}.png - Val/Train Loss and Accuracy.

History

Results from running a model, this is the average accuracy and loss.