Welcome to the TensorFlow Computer Vision Projects repository! This collection is dedicated to exploring and implementing diverse computer vision applications using TensorFlow, ranging from foundational tasks to cutting-edge projects. Here, you'll find projects that dive deep into the power of computer vision, applying neural networks to tasks like image classification, object detection, segmentation, and more.
Each project within this repository will feature:
Custom Architectures & Prebuilt Layers: Combining TensorFlow’s extensive model zoo with custom-designed layers for optimized performance. Advanced Data Augmentation: From traditional techniques to advanced methods like MixUp and CutMix, each project will showcase augmentation methods tailored to different datasets. Custom Training Loops & Metrics: Implementing TensorFlow's custom training capabilities, with hand-crafted loss functions, metrics, and callbacks.
Model & Experiment Tracking: Leveraging tools like TensorBoard and Weights & Biases (W&B) for experiment tracking, dataset versioning, and hyperparameter tuning.
Each project is stored in its own subdirectory with all necessary code, documentation, and configurations. Currently, projects are in progress and will be added here soon. Keep an eye out for:
Malaria Detection 🦠
Object Detection 📦
Image Segmentation 🖍️
Face Recognition 🤖
People Counting
Image Generation
To get started, clone the repository and set up the required dependencies:
Copy code
git clone https://github.com/KaushikML/Computer_Vision_Projects.git
cd Computer_Vision_ProjectsMake sure you have TensorFlow 2.x installed. Additional dependencies for each project will be provided in the respective folders.
🔧 Key Features Data Handling: Efficient preprocessing and data augmentation pipelines. Custom & Transfer Learning Models: Using both custom architectures and transfer learning models. Experiment Tracking with W&B and TensorBoard: Integrated experiment tracking and logging. Hyperparameter Tuning: Fine-tuning models for optimal performance. 💬 Contributions We welcome contributions! If you have an idea for a project or improvement, please feel free to open an issue or submit a pull request.