Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
:credit_card: Detecting the National Identification Cards with Deep Learning (Faster R-CNN)
| Date | Stars |
|---|---|
| 2026-07-24 | 331 |
| 2026-07-25 | 331 |
| 2026-07-28 | 331 |
| 2026-07-30 | 331 |
| 2026-07-31 | 331 |
| 2026-08-06 | 331 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# Detecting the National Identification Cards English | [Türkçe](README.tr.md) You can use this project to extract information DOB (name, surname, date of birth, etc.) on the identity card. To do this, I'm broke down the problem into sub-problems as below: - [this project] Identify Regions of Interest (ROI) containing the required information with deep learning - [this project] Crop the regions identified above - OCR on the identified region of interest This project can do object detection + object classification + multiple object detection all at the same time. **Use case diagram**  Sample id cards | Sample 1 | Sample 2 |Sample 3 | |---|---|---| |  |  | | ## Modernization Notes (2025) This project has been refreshed after 7+ years. It now uses TensorFlow 2 (SavedModel) with optional OCR (EasyOCR). - Removed TensorFlow 1 graph code and TF Object Detection API dependency. No more `object_detection.*` imports. - Loads TF2 SavedModel from `model/saved_model/`; automatically resolves and calls the `serving_default` signature. - `id_card_detection_image.py` CLI flags: - `--image`: Image path (absolute/relative) - `--min_score`: Score threshold (0–1, default 0.60) - `--ocr`: Run OCR (EasyOCR) on the cropped ROI - Lightweight label map parser added; reads ID→name mapping from `data/labelmap.pbtxt`. - Visualization via OpenCV; top-scoring box in green, others red. Cropped ROI is saved as `output_cropped.png`. > Note: SavedModel output keys can vary between models. Defaults expect `detection_boxes`, `detection_scores`, `detection_classes`. Adjust easily if different. ## Setup and Run (Recommended TF2 + OCR) Install either a single platform-specific requirements file (includes TensorFlow), or install TensorFlow separately plus `requirements-modern.txt`. YOLO is optional and can be installed with `ultralytics`. ### Option A) One-shot install (recommended) - Apple Silicon (macOS): ```bash pip3 install -r requirements-macos-apple.txt ``` - Intel macOS / Linux / Windows (CPU): ```bash pip3 install -r requirements-cpu.txt ``` ### Option B) Split install (advanced) Install TensorFlow first, then the rest from `requirements-modern.txt`. ### 1) Install TensorFlow - macOS (Apple Silicon, M-series): ```bash pip3 install tensorflow-macos==2.16.1 tensorflow-metal==1.2.0 ``` - macOS (Intel) or Linux (CPU): ```bash pip3 install tensorflow==2.20.0 ``` - Windows (CPU): ```bash pip3 install tensorflow==2.17.1 ``` > Tip: On Python 3.12, keep pip up to date: `python3 -m pip install --upgrade pip` ### 2) Other dependencies ```bash pip3 install -r requirements-modern.txt ``` ### 3) Run ```bash python3 id_card_detection_image.py --image /absolute/or/relative/path.jpg --ocr --min_score 0.6 ``` - Windows example path: `C:\\path\\to\\image.jpg` - Cropped ROI is written as `output_cropped.png` at project root. - With `--ocr`, extracted text lines are printed to the terminal. ### YOLO integration (optional) Install YOLO backend: ```bash pip3 install ultralytics ``` Run with YOLO instead of TF2 (image): ```bash python3 id_card_detection_image.py --image /path/to/img.jpg --yolo_model yolov8n.pt --min_score 0.4 --ocr ``` Run with YOLO (camera) and enable OCR snapshot panel (keys 1–9 to OCR selected crop): ```bash python3 id_card_detection_camera.py --yolo_model yolov8n.pt --min_score 0.4 --ocr ``` Camera window controls: q quit, p pause/resume, s stop camera, b start camera, 1–9 OCR selected snapshot. ## Command-line Arguments Image script (`id_card_detection_image.py`): - `--image` (string): Absolute or relative path to the image file. If omitted, defaults to `test_images/image1.png`. - `--min_score` (float, default 0.60): Minimum confidence score threshold in [0,1] to visualize and crop detections. Lower it (e.g., 0.3–0.5) to see more candidates. - `--ocr` (flag): I
Excerpt of 6,684 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:f7e4f6b557c37625, topic:object-detection, topic:yolo, readme:object detection
matched fp:f7e4f6b557c37625, topic:deep-learning, topic:tensorflow