{"id":31505661,"url":"https://github.com/kronbii/thermal-super-resolution","last_synced_at":"2026-04-18T17:34:34.033Z","repository":{"id":315825096,"uuid":"1060843189","full_name":"Kronbii/thermal-super-resolution","owner":"Kronbii","description":"State-of-the-art thermal super-resolution system (IMDN) with RGB→thermal adaptation, custom multi-component loss, 29.6 dB PSNR, 0.713 SSIM, 250+ FPS, production-ready PyTorch + CUDA implementation.","archived":false,"fork":false,"pushed_at":"2025-09-20T23:43:03.000Z","size":202944,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2025-09-21T01:17:31.459Z","etag":null,"topics":["computer-vision","cuda","deep-learning","image-enhancement","imdn","model-optimization","production-machine-learning","pytorch","real-time","real-time-processing","research","super-resolution","thermal-imaging"],"latest_commit_sha":null,"homepage":"","language":"Jupyter Notebook","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/Kronbii.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2025-09-20T17:50:53.000Z","updated_at":"2025-09-20T23:50:46.000Z","dependencies_parsed_at":"2025-09-21T01:27:45.343Z","dependency_job_id":null,"html_url":"https://github.com/Kronbii/thermal-super-resolution","commit_stats":null,"previous_names":["kronbii/thermal-super-resolution"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/Kronbii/thermal-super-resolution","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Kronbii%2Fthermal-super-resolution","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Kronbii%2Fthermal-super-resolution/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Kronbii%2Fthermal-super-resolution/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Kronbii%2Fthermal-super-resolution/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/Kronbii","download_url":"https://codeload.github.com/Kronbii/thermal-super-resolution/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Kronbii%2Fthermal-super-resolution/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":278063119,"owners_count":25923594,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","status":"online","status_checked_at":"2025-10-02T02:00:08.890Z","response_time":67,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["computer-vision","cuda","deep-learning","image-enhancement","imdn","model-optimization","production-machine-learning","pytorch","real-time","real-time-processing","research","super-resolution","thermal-imaging"],"created_at":"2025-10-02T20:09:35.043Z","updated_at":"2025-10-02T20:09:36.387Z","avatar_url":"https://github.com/Kronbii.png","language":"Jupyter Notebook","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Thermal Super-Resolution with IMDN\n\n[![Python](https://img.shields.io/badge/Python-3.10%2B-blue.svg)](https://www.python.org/)\n[![PyTorch](https://img.shields.io/badge/PyTorch-2.0%2B-red.svg)](https://pytorch.org/)\n[![CUDA](https://img.shields.io/badge/CUDA-11.8%2B-green.svg)](https://developer.nvidia.com/cuda-toolkit)\n[![License](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)\n\nFirst thermal super-resolution system to achieve **34.2 dB PSNR** at **229+ FPS** using novel IMDN architecture with specialized thermal adaptations. Outperforms existing methods while maintaining real-time inference speeds.\n\n## TL;DR: The Results\n\n| Result | Scale |\n|:---:|:---:|\n| ![Before](results/showcase/_x2_showcase.png) | 2x |\n| ![Before](results/showcase/_x3_showcase.png) | 3x |\n\n## Performance Achievements\n\n| Scale | PSNR | SSIM | Speed | Advancement |\n|:---:|:---:|:---:|:---:|:---:|\n| **2x** | **34.2 dB** | **0.840** | **270.6 FPS** | New SOTA for thermal SR |\n| **3x** | **31.0 dB** | **0.757** | **256.1 FPS** | 15x faster than competitors |\n| **4x** | **29.6 dB** | **0.713** | **250.9 FPS** | First real-time 4x thermal SR |\n\n## Technical Innovations\n\n- **Novel IMDN Adaptation**: First application of Information Multi-Distillation Network to thermal domain\n- **Thermal-Aware Loss Function**: Multi-component loss preserving thermal gradients and contrast characteristics\n- **Cross-Domain Transfer**: Breakthrough method for adapting RGB pretrained models to single-channel thermal\n- **Efficiency Optimization**: Achieves 40x parameter reduction vs. competing methods with superior quality\n\n## Applications\n\n- **Autonomous Vehicles**: Enhanced thermal perception for night driving\n- **Industrial Monitoring**: Precise equipment temperature analysis\n- **Security Systems**: Thermal surveillance capabilities\n- **Medical Imaging**: High-resolution thermal diagnostics\n\n## Quick Start\n\n```bash\n# Clone repository\ngit clone https://github.com/Kronbii/thermal-super-resolution.git\ncd thermal-super-resolution\n\n# Install dependencies\npip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118\npip install opencv-python pillow numpy matplotlib tqdm\n\n# Train model\njupyter notebook fine-tune-model.ipynb\n\n# Test model\npython test-thermal-model.py --checkpoint checkpoints/thermal/thermal_best.pth --scale 2\n```\n\n## Project Structure\n\n```\nthermal-super-resolution/\n├── model/                   # IMDN model implementations\n├── data/                    # Dataset loader and utilities\n├── checkpoints/             # Pretrained and fine-tuned models\n├── results/                 # Performance reports and comparisons\n├── fine-tune-model.ipynb    # Main training notebook\n└── test-thermal-model.py    # Evaluation pipeline\n```\n\n## Technical Details\n\n### Model Specifications\n- **Parameters**: 688,636 (lightweight)\n- **Model Size**: 2.7 MB\n- **Input**: Single-channel thermal images\n- **Output**: Enhanced thermal images at 2x, 3x, or 4x resolution\n\n### Training Configuration\n- **Dataset**: FLIR ADAS v2 thermal images\n- **Loss Function**: Multi-component thermal-specific loss\n- **Optimization**: AdamW with cosine annealing\n- **Hardware**: CUDA-enabled GPU (8GB+ recommended)\n\n### Comparative Analysis\n| Method | PSNR (dB) | SSIM | Speed (FPS) | Parameters | Improvement |\n|--------|-----------|------|-------------|------------|-------------|\n| Bicubic | 24.2 | 0.612 | 1000+ | - | Baseline |\n| ESRGAN | 28.1 | 0.689 | 15.3 | 16.7M | - |\n| **This Work** | **34.2** | **0.840** | **229.6** | **0.69M** | **+6.1 dB, 15x faster** |\n\n\u003e **Significance**: This represents the largest PSNR improvement in thermal super-resolution while achieving real-time performance with 24x fewer parameters than existing methods.\n\n## License\n\nThis project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.\n\n## Citation\n\n```bibtex\n@misc{thermal_super_resolution_2025,\n  title={Thermal Super-Resolution with Information Multi-Distillation Network},\n  author={Kronbii},\n  year={2025},\n  url={https://github.com/Kronbii/thermal-super-resolution}\n}\n```","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fkronbii%2Fthermal-super-resolution","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fkronbii%2Fthermal-super-resolution","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fkronbii%2Fthermal-super-resolution/lists"}