Share · Communicate · Innovate

Billion-scale semi-supervised learning for image classification

This paper presents a study of semi-supervised learning with large convolutional networks. We propose a pipeline, based on a teacher/student paradigm, that leverages a large collection of unlabelled images (up to 1 billion). Our main goal is to improve the performance for a given target architecture, like ResNet-50 or ResNext. We provide an extensive analysis of the success factors of our approach, which leads us to formulate some recommendations to produce high-accuracy models for image classification with semi-supervised learning.

View Details

Hackathon: Build a Recipe Search Website

In this post, we present our work for designing a receipe search website during the 1st Techbow Hackathon. The challenge lies in front-end design and good communication with backend and error handling. During the construction, we use bootstrap framework for the Html part, Vue.js + jQuery in the Javascript part and PHP as the backend server. The website is non-comercial and is only for demo use. Our responsive desgin with real-time API helped us win the champion in the hackathon.

View Details Github

Knowledge Aided Consistency for Weakly Supervised Phrase Grounding

Previous methods address weakly supervised grounding by training a grounding system via learning to reconstruct language information contained in input queries from predicted proposals. Instead, we explore the consistency contained in both visual and language modalities, and leverage complementary external knowledge to facilitate weakly supervised grounding.

View Details Github

Query-guided Regression Network with Context Policy for Phrase Grounding

We adopt a spatial regression method to break the performance limit, and introduce reinforcement learning techniques to further leverage semantic context information. We propose a novel Query-guided Regression network with Context policy (QRC Net) which jointly learns a Proposal Generation Network (PGN), a Query-guided Regression Network (QRN) and a Context Policy Network (CPN).

View Details Github

Graduation Roadtrip on Route 66

This is my first blog on travel. I hope to share my experience on the roadtrip, which may provide useful tips for your future plan. I traveled with my roomates and friends for master graduation. We spent 7 days on the Route 66 and went through ~2/3 of USA. This experience is unique and I would like to share our plan, images and my feelings in this post (in Chinese).

View Details