https://github.com/amirzenoozi/yolo-tf2-object-detection
Object Detection on Image and Video Based on YOLO Model
https://github.com/amirzenoozi/yolo-tf2-object-detection
cli coco coco-dataset deep-learning live livestream machine-learning object-detection object-recognition object-segmentation python stream telegram telegram-bot tensorflow tensorflow2 tf2 watchdog yolo yolov3
Last synced: 7 months ago
JSON representation
Object Detection on Image and Video Based on YOLO Model
- Host: GitHub
- URL: https://github.com/amirzenoozi/yolo-tf2-object-detection
- Owner: amirzenoozi
- Created: 2022-06-18T13:18:11.000Z (over 3 years ago)
- Default Branch: main
- Last Pushed: 2022-06-19T11:26:27.000Z (over 3 years ago)
- Last Synced: 2025-01-21T08:24:22.621Z (9 months ago)
- Topics: cli, coco, coco-dataset, deep-learning, live, livestream, machine-learning, object-detection, object-recognition, object-segmentation, python, stream, telegram, telegram-bot, tensorflow, tensorflow2, tf2, watchdog, yolo, yolov3
- Language: Python
- Homepage:
- Size: 28.3 KB
- Stars: 1
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# YOLO Object Detection 🏷️
We Use Pretrained YOLO Model to Detect Objects
## Requierments 📦
```bash
pip install -r requirements.txt
```
## Donwload and Convert pre-trained YOLO-v3 ⏬
```bash
# YOLO V3
- wget https://pjreddie.com/media/files/yolov3.weights -O models/yolov3.weights
- python convert_to_checkpoints.py --weights ./models/yolov3.weights --output ./checkpoints/yolov3.tf
# YOLO-Tiny V3
- wget https://pjreddie.com/media/files/yolov3-tiny.weights -O models/yolov3-tiny.weights
- python convert.py --weights ./models/yolov3-tiny.weights --output ./checkpoints/yolov3-tiny.tf --tiny True
```
## Video Frame Extractor CLI Options 🎞️
```bash
--frame Frame Threshold #default: 1800 (Every Minutes)
--src Video File PATH #default: 'sample.mp4'
```
You Need to Use This Command:
```bash
python frame.py --frame FRAME_TH --src VIDEO_FILE
```
## Live Stream Frame Extractor CLI Options 📺
```bash
--frame Frame Threshold #default: 1800 (Every Minutes)
--src Video File PATH #default: ''
--dir Save Frames Folder Name #default: '' | (Auto-Generate UUID4)
```
You Need to Use This Command:
```bash
python live.py --frame FRAME_TH --src STREAM_LINK --dir FOLDER_NAME
```
if You want to proccess frames during extraction task, you just need to run:
```bash
python watcher.py --src FOLDER_NAME
```
## Object Detection 📋
```bash
--weights Path To .tf File #default: 'model/model.h5'
--classes Path To Classes File #default: './models/coco.names'
--tiny Use Tiny Model Or Not? #default: False
--num_classes Number Of Classes In The Model #default: 80
--size Resize Images To #default: 416
--image Path To Input Image #default: './data/sample.jpg'
--save Save Or Not #default: False
```
Then You Just Need To Run This:
```bash
# Image
python main.py --image PATH_TO_IMAGE
# Video
python video.py --dir PATH_TO_FRAMES_DIR
```
## Features ✨
- [x] Detect Default COCO Classes
- [x] CLI
- [x] Image Files
- [x] Video Files
- [x] Live Stream
- [ ] Support I18N Classes
- [ ] Telegram Bot
- [ ] Rest API
- [ ] Image Support
- [ ] Video Support
- [ ] GIF Support