AnimalActivityRecognition

Overview

Animal Activity Recognition is a deep learning project focused on animal activity recognition using YOLO object detection. The project aims to identify and classify various animal activities, such as running, walking, eating, etc., from video footage or live camera feed.

Note: This project was designed specifically to be used on Sheep. Modify it as required.

Instructions

1. Clone the Repository: Open your terminal or command prompt and clone this repository using the following command:

   git clone https://github.com/perpetualdarkness/AnimalActivityRecognition.git

2. Navigate to the Project Directory: Change the directory to the YoloAnimalActivityRecognition project:

   cd AnimalActivityRecognition
   

3. Install Dependencies: Ensure you have the required dependencies installed. You can set up a virtual environment and install the dependencies from the requirements.txt file:

   python -m venv env
   source env/bin/activate      # On Windows, use: env\Scripts\activate
   pip install -r requirements.txt

Training

To train a new model, follow the TrainModel.ipynb notebook. This notebook is designed to be run in Google Colab to take advantage of GPU acceleration for faster training. Make sure to upload the sample training data located in the data/ directory (../data) to your Google Colab workspace before executing the notebook.

Testing

Prepare Test Data: Place the images or videos you want to test with the pre-trained best.pt model in the Project/Test/ directory.

Run the Model: Execute the runModel.py script to run the model on the test data:

python runModel.py

View Results: The results of the model's predictions will be saved in the Project/Results/ directory.