Fruit & Vegetable Freshness Detection

Real-time system for detecting and classifying fresh vs rotten fruits and vegetables using a two-stage deep learning pipeline. Built as a Minor Project at Thapathali Campus, IOE, Tribhuvan University (2026).

---

Results

Model	Metric	Score
YOLOv8	mAP@50	92.3%
YOLOv8	mAP@50-95	89.5%
YOLOv8	Precision	90.8%
YOLOv8	Recall	87.5%
MobileNetV3	Classification Accuracy	~92%

---

How It Works

Manual inspection of fruits and vegetables is slow, inconsistent, and contributes to food waste. This system automates that process in real time.

Two models work together in a single pipeline:

Input (Image / Video / Webcam)
↓
YOLOv8 - detects and localizes each fruit/vegetable
↓
MobileNetV3 - classifies each detected object as Fresh or Rotten
↓
Output with bounding boxes + freshness labels

• YOLOv8 handles object detection — finds where the fruits are in the frame
• MobileNetV3 handles classification — evaluates color, texture, and surface patterns to determine freshness
• Both run simultaneously, enabling real-time multi-object assessment in a single frame

---

Project Structure

Freshness-Detection-Computer_Vision/
│
├── models/              # Trained YOLOv8 and MobileNetV3 model weights
├── utils/               # Helper functions for preprocessing and inference
├── Testing/             # Test images and validation scripts
├── output/              # Sample output results
├── code.ipynb           # Main training and experimentation notebook
├── train_demo1.ipynb    # Model training walkthrough
├── final_pipeline.py    # End-to-end inference pipeline
└── testing_model_cnn.py # CNN model testing script

---

Tech Stack

• Python
• YOLOv8 (Ultralytics) - object detection
• MobileNetV3 - image classification via transfer learning
• PyTorch - model training
• OpenCV - real-time video processing
• Django + React - web interface

---

Setup & Run

# Clone the repo
git clone https://github.com/aadityaKS1/Freshness-Detection-Computer_Vision-
cd Freshness-Detection-Computer_Vision-

# Install dependencies
pip install -r requirements.txt

# Run the pipeline
python final_pipeline.py

---

Applications

• Agriculture - quality control at harvest
• Retail & supermarkets - automated shelf inspection
• Supply chain - freshness monitoring during transit

---

Team

• Aaditya Kumar Singh (THA079BCT001)
• Anish Kumar Singh (THA079BCT004)
• Jeewan Bhatt (THA079BCT015)
• Ruby Kumari Sah (THA079BCT034)

Supervised by Er. Shanta Maharjan, Dept. of Electronics & Computer Engineering, Thapathali Campus, IOE

---

Also add these to the repo (if not done yet)

• Add requirements.txt — list all pip packages used
• Add a screenshot or demo GIF in the output/ folder and reference it here
• Add topics on GitHub: computer-vision, yolov8, object-detection, deep-learning, python, mobilenetv3