Multi-Modal Semantic Image Retrieval Demo

A joint multi-modal space for images and words with semantic structure has been learnt from Social Media data in a self-supervised way. This demo lets explore that space, using words and images as queries, and allowing to perform arithmetics between them.
Go to the demo

SetaMind

An Android App that, given a photo of a mushroom, recogonizes its species. Using it is very simple: Take a photo of the mushroom with your phone and SetaMind will recognize the specie and will provide information about it, being able to recognize 24 different species. It uses a CNN that works locally in the phone.
SetaMind is available in Google Play.
Blog post

FaceFCN

A Caffe Fully Convolutional Network for Face and Hair segmentation. Adapted from FCNs for Semantic Segmentation by Long and Shelhamer.
Blog post

TextFCN

A Caffe Fully Convolutional Network that detects any kind of text and generates pixel-level heatmaps. Adapted from FCNs for Semantic Segmentation by Long and Shelhamer. This came out form my MS’s thesis, and lead to ICPR workshop and Pattern Recognition Letters Journal publications.
TextFCN on Github