🌱 Issue Proposal: Plant Seedlings Classification using Deep Learning
Project Title:
Automated Plant Seedlings Classification with CNN Architectures
1. Project Description
Agriculture is one of the most important pillars for human survival and economic growth, particularly in developing countries. With the growing demand for food and increasing challenges posed by climate change, the agricultural sector needs automation to boost productivity while minimizing costs.
This project focuses on automated plant seedling classification using Convolutional Neural Networks (CNNs), leveraging a Kaggle dataset of ~5,000 images across 12 plant species. By applying multiple CNN architectures (ResNet, AlexNet, VGG, Inception, MobileNet, SqueezeNet, DenseNet), the project demonstrates the power of deep learning in weed detection, crop monitoring, and precision agriculture.
The trained model achieved up to 99.48% accuracy on the test set, proving its feasibility as a scalable agricultural AI solution.
2. Problem Statement
Traditional methods for weed recognition and plant classification face challenges of scalability, accuracy, and cost-effectiveness. Farmers require reliable solutions for automated plant identification to reduce manual labor, improve crop yield, and adopt sustainable farming practices.
The current notebook provides a foundation but lacks:
- A modular, production-ready pipeline
- Deployment strategy (interactive app or API)
- Beginner-friendly contribution points
- Comparative visualizations and thorough documentation
3. Proposed Solution / Implementation Plan
We will enhance and extend the current notebook into a community-friendly open-source project with the following phases:
Phase 1: Data Handling & Preprocessing (Beginner-Friendly)
Phase 2: Model Training & Evaluation (Intermediate)
Phase 3: Deployment & Application (Advanced)
4. Tech Stack
- Languages: Python
- Deep Learning Frameworks: TensorFlow, Keras, PyTorch
- Libraries: Pandas, NumPy, Matplotlib, Seaborn, Scikit-learn
- Deployment: Streamlit / Gradio
- Version Control: Git, GitHub
5. Expected Impact
- 📘 Learning Resource: Helps contributors understand CNNs through hands-on experimentation.
- 🌍 Practical Application: Demonstrates how AI can support precision agriculture.
- 🤝 Open-Source Friendly: Provides structured contribution tasks for beginners to advanced participants.
- 🚀 Future-Ready: Can be extended to support multi-crop classification, disease detection, or drone integration.
6. Contribution Roadmap / Labels
good first issue → Data preprocessing, documentation.intermediate → Model training & evaluation.advanced → Deployment & application building.
7. Additional Notes
👉 Proposed by: @keshripritesh
👉 Related to: GSSoC'25
🌱 Issue Proposal: Plant Seedlings Classification using Deep Learning
Project Title:
Automated Plant Seedlings Classification with CNN Architectures
1. Project Description
Agriculture is one of the most important pillars for human survival and economic growth, particularly in developing countries. With the growing demand for food and increasing challenges posed by climate change, the agricultural sector needs automation to boost productivity while minimizing costs.
This project focuses on automated plant seedling classification using Convolutional Neural Networks (CNNs), leveraging a Kaggle dataset of ~5,000 images across 12 plant species. By applying multiple CNN architectures (ResNet, AlexNet, VGG, Inception, MobileNet, SqueezeNet, DenseNet), the project demonstrates the power of deep learning in weed detection, crop monitoring, and precision agriculture.
The trained model achieved up to 99.48% accuracy on the test set, proving its feasibility as a scalable agricultural AI solution.
2. Problem Statement
Traditional methods for weed recognition and plant classification face challenges of scalability, accuracy, and cost-effectiveness. Farmers require reliable solutions for automated plant identification to reduce manual labor, improve crop yield, and adopt sustainable farming practices.
The current notebook provides a foundation but lacks:
3. Proposed Solution / Implementation Plan
We will enhance and extend the current notebook into a community-friendly open-source project with the following phases:
Phase 1: Data Handling & Preprocessing (Beginner-Friendly)
Phase 2: Model Training & Evaluation (Intermediate)
Phase 3: Deployment & Application (Advanced)
Refactor notebook into modular Python scripts and functions.
Develop an interactive web dashboard using Streamlit or Gradio:
Package as a Dockerized API for scalability (optional).
Write comprehensive documentation and tutorials.
4. Tech Stack
5. Expected Impact
6. Contribution Roadmap / Labels
good first issue→ Data preprocessing, documentation.intermediate→ Model training & evaluation.advanced→ Deployment & application building.7. Additional Notes
👉 Proposed by: @keshripritesh
👉 Related to: GSSoC'25