labelImg VS Tesseract.js

In addition to the above comments, check out https://github.com/heartexlabs/labelImg

For setting up object detection data I also really like LabelImg: https://github.com/heartexlabs/labelImg

Our case is a bit specific, but we have had a great experience working with a simple tool called LabelImg (https://github.com/tzutalin/labelImg). The source code is trivial, and you can make some adjustments (line colours, box fill, etc.) before putting it into use with people.

I've used CVAT and LabelImg and they both worked great

Labeling images: depending on the tool, you must label the images in PASCAL VOC format for TF or YOLO for Darknet. You can use programs like labelImg, VIA, VoTT (no longer maintain), among others.

there are loads of labeling tools out there e.g. https://github.com/tzutalin/labelImg

I recently used labelImg. It's free, easy to use and has a simple interface. If you don't want to install nothing you can use VGG Image Annotator because it runs directly in a web browser.

I've never used it - but Tesseract.js looks interesting. Perhaps you can use it as a foundation on which to build something like the AWS tool (which seems to work by overlaying an SVG over the top of the image, then manipulates that SVG when user clicks on a results tag to get the word/phrase's highlight box to display)?

I just played with this for a second https://tesseract.projectnaptha.com/ It seemed to work OK as long as the image was high enough resolution even with a fairly busy background. Took maybe 5-10 seconds to run on a Sea Spawn stat block.

TesseractJS - This is a pure JS port of Tesseract OCR which is a popular Optical Character Recognition engine.

Extract Text from the images and converting them to txt file using Tesseract.js

Have you tried this version?

The total size of the download seems to be 3-4MB (based on https://github.com/naptha/tesseract.js/blob/master/docs/loca...), which is less than I expected.