AI-Based Fruit Freshness & Expiry Prediction Using Images

Job ID: 39750521

Budget: ₹1,500 – ₹12,500 INR

I am a final-year B.Tech CSE student.
I need help in building my academic project:
AI-Based Fruit Freshness and Expiry Prediction Using Images.

This is for educational purposes only, not a commercial system. I only need a working prototype that I can demonstrate in my final-year viva/presentation.


---

Project Objectives

1. Build a Deep Learning Model (CNN / YOLO + CNN) that can:

Classify fruit images into Fresh, Ripe, Overripe, Spoiling, Spoiled.

Predict the number of days remaining before expiry.



2. Create a simple GUI (Streamlit preferred) that allows:

Upload of fruit images.

Display of freshness class + expiry prediction.

Highlight spoiled regions (Grad-CAM heatmap optional).



3. Use open datasets (e.g., Fruits-360 from Kaggle) or a small custom dataset.


4. Ensure the project runs on Google Colab (preferred) or my local PC.




---

Project Timeline (Flexible: 1–2 Months)

Week 1 – Dataset & Preprocessing

Collect dataset (Kaggle + custom images).

Preprocess: resize, normalize, augment.

Deliverable: Preprocessed dataset + notebook.


Week 2 – Base CNN Model

Train CNN model for 5 freshness classes.

Evaluate: accuracy, confusion matrix.

Deliverable: Trained model + evaluation results.


Week 3 – Expiry Prediction & Spoilage Visualization

Add expiry prediction (days left).

Implement Grad-CAM or saliency map for spoiled zones.

Deliverable: Notebook with expiry prediction + heatmap.


Week 4 – GUI Integration

Build Streamlit app: upload → classify → show expiry days + spoilage heatmap.

Deliverable: Running GUI demo.


Week 5 – Final Submission Package

Clean, well-commented Python code.

Trained model weights.

Mini project documentation (15–20 pages).

Step-by-step setup instructions.

Deliverable: Final ZIP with everything.


Note: I am flexible if it takes 6–8 weeks, but weekly updates are required.


---

Deliverables (Clear List)

1. Python source code (TensorFlow / PyTorch, OpenCV).


2. Preprocessed dataset (or dataset link).


3. Trained model weights.


4. Streamlit GUI app.


5. Mini documentation/report (Word/PDF).


6. Setup guide (to run on Colab / local PC).




---

Skills Required

Python

Machine Learning / Deep Learning

CNN (TensorFlow / PyTorch)

Computer Vision (OpenCV)

Streamlit / Flask



---

Budget

Platform Range: ₹1,500 – ₹12,500 INR

Actual Student Budget: Around ₹4,000 – ₹6,000 INR

Payment will be released milestone by milestone only, after demo proof.



---

Important Notes

This is a final-year academic project, not a professional/commercial system.

I only require a working prototype with reasonable accuracy and GUI.

Milestone payments only (no upfront full payment).

Weekly updates (Colab link / screenshots) are mandatory.