Achieving high accuracy for a image processing model can sometimes be challenging. For these scenarios annotating and then segmentation works best to achieve high accuracy. I was trying to train dermanet dataset with high accuracy but with keras and model training techniques alone I could achieve maximum accuracy of 77%. Then after annotating the images the accuracy shooted upto 92%. CVAT by Oracle is a very popular image labelling tool but they are stopping the services. LabelImg is another great tool and everyone should know about it I think. Follow the blog to know more.
The easiest way to download and install LabelImg is via pip, and it assumes you’re running Python3.
Simply run the following in your command line: `pip3 install
labelImg. Then, launch LabelImg by typing
labelImg in your command line prompt.
Or simply clone this git repo and run labelimg.py https://github.com/heartexlabs/labelImg
LabelImg supports labelling in VOC XML or YOLO text file format.
We strongly recommend you use the default VOC XML format for creating labels. Thanks to ImageNet, VOC XML is a more universal standard as it relates to object detection whereas various YOLO implementations have slightly different text file formats.
Moreover, you can always easily convert from VOC XML to any other format using Roboflow, like VOC XML to COCO JSON.
Open your desired set of images by selecting “Open Dir” on the left-hand side of LabelImg
To initiate a label, type w, and draw the intended label. Then, type ctrl (or command) S to save the label. Type d to go to the next image (and a to go back an image).
When labeling images, there are a few tips to bear in mind. Note that these tips are generally true, but can alter based on the context of your problem. Be sure to read our full deep dive on labeling best practices.