Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/matlab-deep-learning/object-detection-using-pretrained-yolo-v2

YOLO v2 prediction and training in MATLAB for Object Detection with darknet19 & tinyYOLOv2 base networks
https://github.com/matlab-deep-learning/object-detection-using-pretrained-yolo-v2

computer-vision deep-learning image-processing matlab matlab-deep-learning object-detection pretrained-models transfer-learning yolo yolov2

Last synced: 1 day ago
JSON representation

YOLO v2 prediction and training in MATLAB for Object Detection with darknet19 & tinyYOLOv2 base networks

Awesome Lists containing this project

README 

        
# Pretrained YOLO v2 For Object Detection



This repository implements pretrained YOLO v2 [1] object detectors in MATLAB. [![Open in MATLAB Online](https://www.mathworks.com/images/responsive/global/open-in-matlab-online.svg)](https://matlab.mathworks.com/open/github/v1?repo=matlab-deep-learning/Object-Detection-Using-Pretrained-YOLO-v2)



**Creator**: MathWorks Development



## Requirements  

- MATLAB® R2020a or later

- Deep Learning Toolbox™

- Computer Vision Toolbox™

- Computer Vision Toolbox™ Model for YOLO v2 Object Detection



Note: Previous MATLAB® release users can use [this](https://github.com/matlab-deep-learning/Object-Detection-Using-Pretrained-YOLO-v2/tree/previous) branch to download the pretrained models.



## Getting Started

[Getting Started with YOLO v2](https://in.mathworks.com/help/vision/ug/getting-started-with-yolo-v2.html)



### Detect Objects Using Pretrained YOLO v2

Use to code below to perform detection on an example image using the pretrained model.



Note: This functionality requires Deep Learning Toolbox™ and the Computer Vision Toolbox™ for YOLO v2 Object Detection. You can install the Computer Vision Toolbox for YOLO v2 Object Detection from Add-On Explorer. For more information about installing add-ons, see [Get and Manage Add-Ons](https://in.mathworks.com/help/matlab/matlab_env/get-add-ons.html).



```

% Load pretrained detector

modelName = 'tiny-yolov2-coco';

detector = yolov2ObjectDetector(name);



% Read test image.

img = imread('sherlock.jpg');



% Detect objects in the test image.

[boxes, scores, labels] = detect(detector, img);



% Visualize detection results.

img = insertObjectAnnotation(img, 'rectangle', bboxes, scores);

figure, imshow(img)

```

![alt text](images/results.jpg?raw=true)



### Choosing a Pretrained YOLO v2 Object Detector

You can choose the ideal YOLO v2 object detector for your application based on the below table:



| Model | mAP | Size (MB) | Classes | Speed in Frames Per Second (FPS) |

| ------ | ------ |  ------ | ------ | ------ |

| Darknet19-COCO | 28.7 | 181 | [coco class names](+helper/coco-classes.txt) | 17.8 |

| Tiny-YOLO_v2-COCO | 10.5 | 40 | [coco class names](+helper/coco-classes.txt) | 32 |



- Performance (in FPS) is measured on a TITAN-XP machine using:

    - 608x608 image for Darknet19-COCO.

    - 416x416 image for Tiny-YOLO_v2-COCO.

- mAP for models trained on the COCO dataset is computed as average over IoU of .5:.95.



### Train Custom YOLO v2 Detector Using Transfer Learning

To train a YOLO v2 object detection network on a labeled data set, use the [trainYOLOv2ObjectDetector](https://in.mathworks.com/help/vision/ref/trainyolov2objectdetector.html) function. You must specify the class names for the data set you use to train the network. Then, train an untrained or pretrained network by using the [trainYOLOv2ObjectDetector](https://in.mathworks.com/help/vision/ref/trainyolov2objectdetector.html) function. The training function returns the trained network as a [yolov2ObjectDetector](https://in.mathworks.com/help/vision/ref/yolov2objectdetector.html) object.



For more information about training a YOLO v2 object detector, see [Object Detection using YOLO v2 Deep Learning Example](https://www.mathworks.com/help/vision/ug/train-an-object-detector-using-you-only-look-once.html).



## Code Generation

Code generation enables you to generate code and deploy YOLO v2 on multiple embedded platforms. For more information about generating CUDA® code using the YOLO v2 object detector see [Code Generation for Object Detection by Using YOLO v2](https://www.mathworks.com/help//deeplearning/ug/code-generation-for-object-detection-using-yolo-v2.html)



## YOLO v2 Algorithm Details

YOLO v2 is a popular single stage object detectors that performs detection and classification using CNNs. The YOLO v2 network is composed of a backbone feature extraction network and a detection head for the localization of objects in an image. For more information about YOLO v2, see [Getting Started with YOLO v2](https://www.mathworks.com/help/vision/ug/getting-started-with-yolo-v2.html). 



![alt text](images/yolo_model.png?raw=true) 



## References

[1] Redmon, Joseph, and Ali Farhadi. "YOLO9000: better, faster, stronger." Proceedings of the IEEE conference on computer vision and pattern recognition. 2017.



[2] Lin, T., et al. "Microsoft COCO: Common objects in context. arXiv 2014." arXiv preprint arXiv:1405.0312 (2014).



[3] The PASCAL Visual Object Classes Challenge: A Retrospective Everingham, M., Eslami, S. M. A., Van Gool, L., Williams, C. K. I., Winn, J. and Zisserman, A. International Journal of Computer Vision, 111(1), 98-136, 2015.



Copyright 2021 - 2024 The MathWorks, Inc.