Yolov3 For Custum Objects

This repository illustrates the steps for training YOLOv3 and YOLOv3-tiny to detect fire in images and videos. First, a fire dataset of labeled images is collected from the internet. The images with their annotations have been prepared and converted into YOLO format and put into one folder to gather all the data. Therefore, the data folder contains images ('*jpg') and their associated annotations files ('.txt') with the same name.