This page lists projects I have participated and corresponding result.

Superpixel-enhanced pairiwise Conditional Random Field for Semantic Segmentation

The first pairwise fully-connected CRF model that incorporate superpixel cues. Our model has better accuracy and speed compared with corresponding conventional superpixel-based higher-order CRFs.

ROI detection in large histopathological medical images.

Proposed and implemented a coarse-to-fine multi-level unsupervised segmentation algorithm for large images.

Data Processing Basic Software.

Realized basic machine learning algorithm using C++, including linear regression, logistic regression, neural network, SVM, hierarchical clustering, apriori, k-means and so on. It has a MFC interface to set up parameters.