annotationsobject-detectionobject-recognition

fast image annotation for large dataset


I have a dataset related to plant disease. the dataset has 70k images and 38 classes dataset and I want to annotate images with a bounding box. I try for annotation LabelImg and other annotation tools but all of them are so time-consuming. is there any other technique for image annotation which is fast, simplest, and automatic or can I do with programming?

please help me out to solve this problem.


Solution

  • I used hasty.ai on a recent project. Their tool has an AI which observes you while you label. After learning on a few images only it suggests annotations which you can accept or correct (and by that retrain the model). Once the AI makes good annotations, you can batch process all your remaining images. It allowed me to annotate a huge dataset within a few days only.

    You can use it for free in the beginning! I switched to pro, then, as it was worth the money and the price is reasonable.