Push the limit for image classification using MIT-IBM ObjectNet dataset




Abstract:   Deep Learning based mage classification performance on the ImageNet dataset has now almost saturated.    In this project we will research advanced deep learning algorithms for visual object recognition on the new and much more challenging data set of MIT-IBM ObjectNet.    Outcomes of this project will lead to major progress in computer vision and machine learning field.

Workload:  a 12-unit project, to be completed in two consecutive semesters. 

ContactHongdong Li

Updated:  1 June 2019/Responsible Officer:  Dean, CECS/Page Contact:  CECS Marketing