farmers-efficiency-jetson-nano

This project aims to enhance farming efficiency using computer vision and deep learning techniques implemented on a Jetson Nano device. The project consists of three main components: weed detection using YOLOv4 and YOLOv5, and lane detection using Hough Transform.

Project Components

1. YOLOv4 Weed Detection

Located in yolov4-Weed-Detection/, this component uses YOLOv4 for real-time weed detection in agricultural fields.

• Features:
  • Image-based weed detection (detect_image.py)
  • Video-based weed detection (detect_video.py)
  • Pre-trained weights included (crop_weed_detection.weights)
  • Configuration file for model architecture (crop_weed.cfg)
• Note: This implementation runs on CPU as it doesn't support GPU acceleration on Jetson Nano

2. YOLOv5 Weed Detection

Located in yolov5-Weed-Detection/, this is an improved version using YOLOv5 for weed detection with GPU acceleration support.

• Features:
  • Real-time detection capabilities (detectRT.ipynb)
  • ONNX model support for optimized inference (detect_using_onnx.ipynb)
  • Image and video detection scripts
  • Pre-trained ONNX model included (best.onnx)
  • TensorRT optimization for GPU acceleration on Jetson Nano
  • Significantly improved inference speed compared to YOLOv4 implementation
  • Hardware-accelerated inference leveraging Jetson Nano's GPU capabilities

3. Lane Detection

Located in Hough-transform-laneDetection/, this component implements lane detection using Hough Transform for agricultural vehicle navigation.

• Features:
  • Lane detection in agricultural fields
  • Jupyter notebook implementation for easy experimentation

Setup Instructions

Prerequisites

• Jetson Nano device
• Python 3.x
• CUDA support (for GPU acceleration)
• Necessary Libraries such as: OpenCV, TensorRT (for YOLOv5 GPU optimization), Matplotlib and NumPy

Installation

1. Clone the repository:

git clone [repository-url]
cd farmers-efficiency-jetson-nano

2. Setup the environment with required dependencies on jetson nano.

Usage

YOLOv4 Weed Detection

# For image detection
python detect_image.py 

# For video detection
python detect_video.py 

YOLOv5 Weed Detection

# For real-time detection with GPU acceleration
python detectRT.ipynb

# For ONNX model inference with TensorRT optimization
python detect_using_onnx.ipynb

Lane Detection

Open and run the Jupyter notebook in the Hough-transform-laneDetection directory.

Contributors

• Soham Kukreti
• Yuvraj Rathi
• Satyam Rathi