NJUST KMG

School of Computer Science and Engineering,
Nanjing University of Science & Technology

Code

Our Codes are available on our GitHub account njustkmg.

Welcome to use OMML : Multi-Modal learning toolkit based on PaddlePaddle and PyTorch, supporting multiple applications such as multi-modal classification, cross-modal retrieval and image caption.

Baidu paddle developers 《Multimodal learning research and toolkit PaddleMM introduction and application》, [BiLibili],
PaddlePaddle Multi-modal learning toolkit based on PaddlePaddle: PaddleMM, Kaggle Contest Book[PaddleMM],
Our mulan-paddlemm , which we partnered with PaddlePaddle, has now joined the Mulan community. [PaddleMM officially entered the Mulan open source community for incubation] [The 11th TOC meeting of the Mulan Open Source Community in 2022]



Demo

We have developed a website that can identify its authenticity and branding, Ysneaker, the main functions are as follows.

identify

Multimodal Fusion Classification

Trans

Cross-modal Style Transfer

Re

Cross-modal Retrieval



We have developed a website for garbage sorting and detecting new categories of Garbage Classification Website, the main functions are as follows.

Garbage detection

Detection of new class garbage in open scenarios



We have developed a website for generating descriptions and description enhancements for home improvement images Home Description, , the main functions are as follows.

Home re

Enhance the semantic description of home improvement pictures

Home op

Optimize the semantic description of home improvement pictures