Skip to content

Sinha199/Mask-Detection

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

11 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Mask-Detection

The given repo mainly shows how to train custom objects on yolov3.

About YOLO

You only look once (YOLO) is a state-of-the-art, real-time object detection system.The full detail of the working of YOLO and its performance is given in detail in the following link.

YOLO

Training Custom objects

Yolov3 is trained on coco dataset but we can train it for our own custom datasets,detailed information is given on the particular github repo

darknet-custom objects

Mainly four custom files should be added for detecting the desired objects, as I have done for mask detection.

obj.data

obj.names

yolov3_obj.cfg

Important points for training

As per my model I have trained it till 2000 iterations(total 6000 iterations) that took me nearly 6 hours because of GPU constraints.For efficient training store the last trained model and then ,resume the training from the last pretrained weights.This is the link to the 2000th iterations of the model. This can be used for further efficient training, Trained weights

Dataset

This is link to the dataset, images

The folder containes the images and the corresponding text files which contain the annotations of the masked and non-masked images. For annotations of images I have used LabelImg , its a tool graphical image annotation tool.

Overview of Prediction

output