This project focuses on building a deep learning model for detecting weeds in images and highlighting them with green bounding boxes. The model is trained using a labeled dataset containing images of crops and weeds along with bounding box annotations.
Features: Utilizes deep learning techniques for object detection, specifically weed detection. Preprocesses and augments the dataset to enhance model robustness. Implements a custom model architecture with bounding box regression for accurate localization. Provides functionality to draw green boxes around detected weeds in images.
Technologies Used: Python Pytorch Yolo TensorFlow/Keras OpenCV NumPy Google Colab (for training on cloud GPU) D ataset: The dataset used for training contains images of crops and weeds, with bounding box annotations for weeds. The dataset is available on Kaggle.
Usage: Download the dataset and save it to a local directory. Preprocess and augment the dataset using the provided code. Train the weed detection model using TensorFlow/Keras. Evaluate the model's performance and deploy it for weed detection in new images.
Repository Structure: data/: Contains the dataset and preprocessing scripts. models/: Includes the trained model and related files. notebooks/: Jupyter notebooks for data exploration, model training, and evaluation. src/: Source code for the weed detection model and related utilities. README.md: Detailed instructions, project overview, and usage guide.