Projects

IoT‑based UGV Robot for Remote Soil Health Sensing

End‑to‑end autonomous system for precision agriculture: UGV collects soil/env/GPS data and streams to a local base for real‑time analysis and storage.

Raspberry Pi Arduino Nano NRF24 GPS NPK/DHT11 Matplotlib
Code
▶ View details
  • Designed a UGV for soil data collection and reconnaissance.
  • Integrated NPK, temperature, and humidity sensors with GPS tracking.
  • Configured NRF24 wireless link from UGV to base station.
  • Built a Python pipeline to process, visualize, and store CSV sensor streams.
  • Local dashboard for UGV status, sensor readings, and location.
Leaf Disease Detection System

DL system to detect and classify plant leaf diseases; trained on PlantVillage and deployed on an autonomous rover.

PyTorch OpenCV ResNet EfficientNet ViT Flask Tkinter
Code
▶ View details
  • Transfer learning on PlantVillage dataset.
  • Real‑time classification and disease prediction.
  • Deployed on a rover for smart farming.
Multimodal Movie Genre Classification Website

Web app that predicts movie genres using text (overview) + image (poster) features with DL models.

PyTorch Flask HTML NLP CNN CSS
Code Live
▶ View details
  • CNN for posters + NLP for overviews; early/late fusion variants.
  • Flask backend + React frontend for interactive predictions.
  • Ensemble techniques to boost multi‑label accuracy.
Hand Gesture Robot Control

Control a TurtleBot3 robot in Gazebo using hand gestures from your webcam. Real-time gesture recognition with 98% accuracy using SigLIP vision model.

ROS Gazebo Python OpenCV Flask SigLIP TurtleBot3
Code
▶ View details
  • Recognizes 19 different hand gestures with 98% accuracy using SigLIP vision transformer model.
  • Real-time gesture-to-robot control: open palm (forward), OK sign (backward), one finger (left), two fingers (right), fist (stop).
  • Cross-platform camera streaming (Windows/Mac) with Flask server for WSL2 integration.
  • Frame-by-frame analysis with 10-frame smoothing for stable gesture recognition.
  • Publishes velocity commands to TurtleBot3 via ROS /cmd_vel topic.
Crop Yield Prediction & Optimization

Data pipelines + ML to predict yields and key drivers across regions; dashboard for insights.

scikit‑learn TensorFlow/Keras Pandas Airflow SQL
Code
▶ View details
  • Collected, cleaned, and fused historical crop, weather, soil, and irrigation data.
  • Benchmarked models to forecast yields and analyze feature importances.
  • Built an interactive dashboard to explore predicted vs. actual outcomes.
Automated CNC Potting Machine

An open-source, scalable precision farming system to automate crop manufacturing and nurturing.

Arduino Mega 2560 Soil Moisture Sensors IR Sensors NEMA 17 A4988 Drivers Open-Source
Code
▶ View details
  • Orchestrates farming functions with an Arduino Mega 2560 microcontroller.
  • Monitors soil conditions and tray positions in real-time using soil moisture and IR sensors.
  • Enables intelligent, data-driven decisions for plowing, seeding, and watering.
  • Uses NEMA 17 stepper motorsand A4988 drivers for precise tool control.
  • Features a flexible, scalable, and open-source design adaptable to various crops and pot sizes.