Welcome! This repository is a collection of projects across core Computer Vision tasks, implemented with PyTorch and other open-source libraries. It includes beginner-friendly projects (like CNN from scratch) and extends to advanced topics such as segmentation, image captioning, VQA, and 3D vision.
| Task | Description | Status |
|---|---|---|
| CNN From Scratch | Build a convolutional neural network manually with PyTorch | ✅ |
| Image Classification | Binary, Multiclass, and Multilabel classification tasks | ✅ |
| Object Detection | YOLO, SSD, Faster R-CNN based models | 🛠️ |
| Image Segmentation | Semantic, instance, panoptic segmentation | 🛠️ |
| Pose Estimation | Detect keypoints of humans (2D/3D) | ⏳ |
| Depth Estimation | Monocular and stereo image depth prediction | ⏳ |
| Image Captioning | Generate image descriptions using RNNs/Transformers | ⏳ |
| VQA | Visual Question Answering combining NLP + CV | ⏳ |
| Image Translation | Style transfer, colorization, inpainting | ⏳ |
| Anomaly Detection | Defect detection using unsupervised/autoencoders | ⏳ |
| Image Retrieval | Find similar images using embeddings | ⏳ |
| OCR | Detect & recognize printed/handwritten text | ⏳ |
| 3D Vision | 3D reconstruction, point clouds | ⏳ |
| Video Analysis | Tracking, action recognition, temporal detection | ⏳ |
| Face Analysis | Detection, recognition, landmarks, deepfake | ⏳ |
🔄 Projects will be updated progressively. Feel free to star the repo for updates!
- Languages: Python 3.x
- Libraries: PyTorch, OpenCV, Torchvision, Numpy, Matplotlib
- Tools: Jupyter, Streamlit (for apps), Gradio (for demos)
Each project lives in its own folder with:
- Code (
.ipynb,.py) README.mddescribing the task, dataset, and method- Optional: demo app (Streamlit, Gradio)
To add a new project:
mkdir new-task-folder
# Add code, dataset link, results, and update this README
## 🤝 Contributions
Feel free to fork this repo, raise issues, or suggest new CV projects.
---
## 📬 Contact
- LinkedIn: [Your LinkedIn link]
- Email: [Your email]
---
> ⭐ If you find this useful, give it a star!