Organize Media AI

This Python script organizes images and videos by date and event, renames them systematically, detects duplicates (exact and near), and uses AI to label images.

📁 Features

• 🗃️ Organizes files into /destination/YYYY/MM-DD/EventName
• 🧠 Labels images using a pretrained ResNet-50 model
• 🔒 Detects exact duplicates using SHA-256
• 🧬 Detects near-duplicates (images & videos) using perceptual hashing
• 📝 Stores rename counters in organize_config.json
• 🎥 Uses multiple video frames to improve duplicate accuracy
• 🛠️ Dry run mode to test before going live
• 🚀 Ski treating same images on different batches storing the list of processed hashes

🛠️ Requirements

pip install pillow torchvision timm exifread imagehash opencv-python pyheif

On macOS, you may also need:

brew install libheif

🚀 Usage

Put all your unsorted media in the /source folder.

Run the script:

python organize_media.py

or run

make run

Sorted files will be moved to /destination.

🧾 Output Structure

/destination/
├── 2024-
│   └── 03-10/
│       └── Birthday/
│           ├── pict00001_dog.jpg
│           └── vid00001_party.mp4
├── duplicated/
├── near_duplicated/
└── noevaluate/

📦 Configuration The script stores its counters in organize_config.json, so it can resume where it left off on the next run.

---

🔧 Configuration File

After the first run, organize_config.json is created:

{
  "pic_counter": 34,
  "vid_counter": 12
}

This ensures that counters continue across multiple runs.

---

🧠 AI Classification

Images are classified with ResNet-50 using PyTorch + TIMM. The label is added to the filename automatically.

Example:

Original: IMG_1234.JPG
Renamed:  pict00012_golden_retriever.jpg

---

🧬 Duplicate Detection

• Exact duplicates: Detected with SHA-256 file hash
• Near-duplicates:
  • Images: imagehash.phash
  • Videos: Sample 5 frames → compute average perceptual hash

These are moved to:

/destination/exact_duplicated/
/destination/near_duplicated/

---

✅ How to Use

1. Drop unsorted media into /source
2. Run the script:

python organize_media.py

3. Organized files will be moved to /destination

---

🧪 Tested On

• Python 3.10+
• macOS and Linux
• NVIDIA CUDA GPU (optional for faster AI classification)

---

📜 License

MIT License

---

🙌 Contributions

Feel free to fork and contribute! Ideas, PRs, and issues are welcome.

---

Happy organizing! 🧹📁📸

---

🧑‍💻 Author

Developed by Nynor-code.