README.md

metadata

language: en
license: mit
tags:
  - image-classification
  - pytorch
  - cnn
  - fruit-classifier
pipeline_tag: image-classification

🍎 Fruit Classifier

Custom CNN trained from scratch in PyTorch to classify 10 fruits from images. Achieves 61.3% validation accuracy on just 230 training images.

Usage

from predict import load_model, predict

model = load_model("fc_model_weights.pth")
fruit_name, confidence = predict(model, "your_image.jpg")
print(f"Predicted: {fruit_name} ({(100*confidence):>0.1f}%)")

Supported Fruits

Apple, Banana, Avocado, Cherry, Kiwi, Mango, Orange, Pineapple, Strawberries, Watermelon

Files

• fc_model.pth — full model for inference
• fc_model_weights.pth — weights for resuming training
• fruit_benefits.json — nutritional knowledge base
• predict.py — inference utility

Full Project

Training pipeline and RAG workflow available on GitHub: GitHub Repository