https://github.com/qubitpi/character-detection
https://github.com/qubitpi/character-detection
Last synced: 30 days ago
JSON representation
- Host: GitHub
- URL: https://github.com/qubitpi/character-detection
- Owner: QubitPi
- Created: 2023-12-03T09:14:05.000Z (over 1 year ago)
- Default Branch: master
- Last Pushed: 2023-12-03T09:59:46.000Z (over 1 year ago)
- Last Synced: 2025-02-07T14:15:39.417Z (3 months ago)
- Language: Python
- Size: 414 KB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
* [Basic - Does not save video output, but image output is saved](./basic.py)
- `python basic.py -v in.MP4 -o out.avi` ([OpenCV for video containers supports only the **avi** extension](https://stackoverflow.com/a/48484222/14312712) and size of Writer have to matched with the frame size of video)
- `python basic.py -i image.png -o out.png`
- https://data-flair.training/blogs/python-project-real-time-human-detection-counting/
- [Histograms of Oriented Gradients for Human Detection](./Histogram%20of%20Oriented%20Gradient%20Descriptor.pdf)
- [HOG wikipedia](https://en.wikipedia.org/wiki/Histogram_of_oriented_gradients)
- Detects "multiple persons" in each frame. In order to track the "right" person, a simple technique would be to pick
up the nearest person in the next frame due to how a person moves.
- **_This is essentially an object detection not people/character detection. So we might not use this script_**
Then [next](https://medium.com/@madhawavidanapathirana/https-medium-com-madhawavidanapathirana-real-time-human-detection-in-computer-vision-part-1-2acb851f4e55):
Early approaches for Human Detection
------------------------------------
### Haar Cascades for Human Detection
Haar feature based approach for object detection is proposed by Paul Viola and Michael Jones in their paper
[Rapid Object Detection using a Boosted Cascade of Simple Features](./Rapid%20Object%20Detection%20using%20a%20Boosted%20Cascade%20of%20Simple%20Features.pdf) published in 2001. This
approach is widely used for Face Detection.
**OpenCV** includes inbuilt functionality to provide Haar cascade based object detection. Pre-trained models provided by OpenCV for "**Full Body Detection**", "**Upper Body Detection**" and "**Lower Body Detection**" are available
[here](https://github.com/opencv/opencv/tree/master/data/haarcascades).
Modern approaches for Human Detection
-------------------------------------
### Human Detection using Tensorflow Object Detection API
#### Setting up a Basic Human Detector
1. Make sure **Open CV** 3.0 (or above) and **Tensorflow** 2.0 (or above) are installed. It is recommended to use the
Tensorflow GPU version when Nvidia GPU is available
2. Download
[faster_rcnn_inception_v2_coco.tar.gz](http://download.tensorflow.org/models/object_detection/faster_rcnn_inception_v2_coco_2018_01_28.tar.gz)
from Tensorflow Detection Model Zoo. Extract the downloaded file. If remote download is not available, here is the
[local alternative](./faster_rcnn_inception_v2_coco_2018_01_28.tar.gz) that has been downloaded from there.
3. `python3 tensorflow_basic_human_detector.py -m path/to/faster_rcnn_inception_v2_coco_2018_01_28/frozen_inference_graph.pb -v test.mp4 -o out`
and observe the output on screen. (Press 'Q' to exit). May adjust the threshold parameter to improve the results.