Bin Jin

Research Interests

  1. Computer Vision
  2. Image Processing
  3. Deep Learning
  4. Computational Aesthetics

Education

  • 2012 – now: PhD. student in IVRL, EPFL.
  • 2008 – 2012: B.Sc. in Communcation Engineering, Shanghai Jiaotong University, China.

Research projects

Image Aesthetics:

In this project, we aim at predicting the aesthetic quality of images using Convolutional Neural Network.

Face Saliency:

In this project, we aim at modeling the different importance of faces in images. We propose a 2D Gausiian model for face importance and build a face saliency algorithm accordingly.

Semantic Segmentation:

In this project, we aim at training neural networks for semantic segmentation, using web images in a weakly-supervised manner. We propose a three-stage training pipeline to train segmentation network using web images. Our algorithm produces state-of-the-art segmentation results.

Publications

[epfl:infoscienceFilteredBox]