Computer Vision

A collection of computer vision models and projects, specifically covering image classification and object detection.

Installation and Usage

Setup your environment and install the required dependencies as follows:

1. Clone the Repository:

git clone https://github.com/fraserlove/computer-vision.git
cd computer-vision

2. Create a Python Virtual Environment:

python -m venv .venv
source .venv/bin/activate

3. Install Dependencies via PIP:

pip install -r requirements.txt

4. Run a Jupyter Notebook server

jupyter notebook

Included Models

• Image Classification
  • Binary Classifier
  • Multi-label Classifier (Feed-Forward Neural Network)
  • Multi-label Classifier (CNN)
  • Multi-label Classifier (Transfer Learning with Pre-Trained EfficientNet Model from Keras Applications)
• Object Detection
  • Inference with the TensorFlow Object Detection API and TensorFlow Hub
  • Fine Tuning with the TensorFlow Object Detection API
  • Yolo NAS
  • Yolo v8

A Note on Datasets

Kaggle is used for downloading datasets. Set up an account and generate an API key. Then enter the following, replacing USERNAME and API_KEY with their values.

mkdir ~/.kaggle
echo 'api_token = {"username":USERNAME,"key":API_KEY}' >> ~/.kaggle/kaggle.json
chmod 600 ~/.kaggle/kaggle.json