NFT appraisal pricing model using keras, Deep learning using CNN + autoencoder + ImageRegressor to get a good range of your price


- A complex NFT price prediction model using autokeras. - Based on the previous trades, creator(s), platform sold at, volatility. - This model can also used to predict what could be the price of an NFT if you make them and are ready to put it on a platform. - So far rarible, cryptokittes and cryptopunks are learnt fully by alphaclaw and many other nft tokens

AlphaClaw showcase

How it's made

- Using the state of the Art image regression models from SDDs to YOLO to resnets . - Tested each model to pick the most accurate model. - Used the whole of cryptokitties , cryptopunks and some of rarible as a dataset with their past prices + owner profiling. - The app runs on a flask server, the model is a tensorflow model which does the magic. - The front-end is vanilla HTML + CSS and a few more beautifying libraries. (because I am solo and didn't have enough time)

Technologies used

Rarible ProtocolThe GraphUpshot