The only model that I found so far that extract table data with OCR is dots.ocr. Models that came after it have not done a good job. Interesting on testing this new model.

You know it only hit me when I think about tender I think of meat.

Thanks

Thanks.. Sadly I cannot update the url.

There exist too many scenarios where an artist work is copied and published on the NFT marketplaces without their permission. Hence www.nftdetect.com hopes to solve this by searching for one’s work against the major NFT marketplaces.

Great idea!

The matching algorithm currently provides too many false positives. (Ex. use the image from https://opensea.io/assets/0xbc4ca0eda7647a8ab7c2061c2e118a18... , there are 5 potential results, but none had real hits.)

Hi Dropnerd, I tweaked the algorithm to not match similar item but only images with almost exact content. Give it a try.

Less false positives, thanks. I think it's now matching previous owners.

On the product:

I imagine no one actually wants to use the web interface to manually scan on a day to day basis.

My advice is a few cached queries that are really fast that show how this works, custom image upload for people to try out to test if it actually works, and then selling an automated scanning service with alerting for the projects that are looking for copycats.

Thanks dropnerd. I am working on a machine learning feature to filter out false positives internally . Any other suggestions ?

A tale that is foundational and reoccurring.