https://github.com/xueeinstein/darknet-vis

Visualize YOLO feature map in prediction for easily checking your model performance
https://github.com/xueeinstein/darknet-vis

darknet object-detection yolo

Last synced: 8 days ago
JSON representation

Visualize YOLO feature map in prediction for easily checking your model performance

• Host: GitHub
• URL: https://github.com/xueeinstein/darknet-vis
• Owner: xueeinstein
• License: other
• Created: 2017-09-26T11:10:02.000Z (about 8 years ago)
• Default Branch: master
• Last Pushed: 2018-02-12T13:51:49.000Z (over 7 years ago)
• Last Synced: 2025-04-10T04:41:46.917Z (6 months ago)
• Topics: darknet, object-detection, yolo
• Language: C
• Homepage:
• Size: 4.9 MB
• Stars: 72
• Watchers: 2
• Forks: 8
• Open Issues: 5
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
![Darknet Logo](http://pjreddie.com/media/files/darknet-black-small.png)

# Darknet

[Darknet](https://github.com/pjreddie/darknet) is an open source neural network framework written in C and CUDA. It is fast, easy to install, and supports CPU and GPU computation.

For more information see the [Darknet project website](http://pjreddie.com/darknet).

For questions or issues please use the [Google Group](https://groups.google.com/forum/#!forum/darknet).

## Darknet-vis

### Why does visualization matters?

Typically, if you follow the instruction on the YOLO homepage, execute

```

./darknet detector test cfg/coco.data cfg/yolo.cfg yolo.weights data/dog.jpg

```

you get the prediction results:

![yolo-v2 result](data/predictions.jpg)

That looks good, however it doesn't reveal the detector's whole ability due to its fixed confidence threshold which is 0.6 by default.

If you use __darknet-vis__, besides the results image, it also generates objectness visualization and classes visualization.

![yolo-v2 obj confidence](data/yolo_vis_obj.png)

![yolo-v2 classes](data/yolo_vis_class.jpg)

Note that each grid cell corresponds to a unique position on the final feature map. In other words, it represents __receptive field__. As you can see, the YOLOv2 model also detect the trash bin at the right-top corner and the motorbike at the left-top corner with very low confidence. After decrease the threshold to 0.1, the new objectness visualization reveals more details as following

![yolo-v2 obj confidence-2](data/yolo_vis_obj2.jpg)

The trash bin is recognized as a person! Moreover, for small objects or some objects with special aspect ratio, the predicted center of object has a remarkable shift. All of these help you understand your YOLO model better, especially why it gets very low AP score for small objects detection.

### How to use?

Replacing Darknet with Darknet-vis is effortless, you just need to add `visualization=1` in __net__ entry of the cfg file. For example, my config in [yolo.cfg](cfg/yolo.cfg)

#### Python API

I wirte a more handy Darknet API wrapper. You can find the example at [darknet.py](https://github.com/xueeinstein/darknet-vis/blob/master/python/darknet.py). Note that `Darknet.detect_v()` is for this customized visualization.