Reading group Wiki

Machine Learning for Imaging
Computer Laboratory
University of Cambridge

Image
Image processed by Deep dream generator.

About the reading group

I have started a reading group on Machine learning for imaging at the Computer Laboratory (University of Cambridge). Please join us if you are interested or let me know if you wish to be on the mailing list!

It usually takes place on tuesdays from 16.00 to 17.00 in SS28 (corridor of the Rainbow group). Every week we pick up a paper, go over it and finish the reading with a discussion. Ideally, everyone would have read the paper beforehand and someone leads the discussion presenting the paper in an informal manner (without need of slides!). The group would be mostly focused on any machine learning approach applicable to computer vision and computer graphics, but also open to other more theoretical approaches in machine learning.

Please find below the information for each week.

Calendar for the reading group

10th Nov, HDR image reconstruction using CNNs

Dr Rafal Mantiuk will present the paper HDR image reconstruction from a single exposure using deep CNNs published in SIGGRAPH 2017. More information on the project and the paper here.

17th Nov, CNNs for object detection from videos

Dr Maria Pérez (i.e., me) will present the paper T-CNN: Tubelets with Convolutional Neural Networks for Object Detection from Videos published in CVPR 2016. Paper accessible here.

24th Nov, Moving object detection using CNNs

Dr Bihao Wang will present the paper MODNet: Moving Object Detection Network with Motion and Appearance for Autonomous Driving published in CVPR 2017. Paper accessible here. Please note that this session of the reading group will take place from 14.00 to 15.00 because of overlapping with other talks.

1st Dec, Object segmentation using mask region-based CNNs

Simeon Spasov will present the paper Mask R-CNN published in ICCV 2017. Paper accessible here. Code accessible here. Please note that this session of the reading group will take place from 14.00 to 15.00.

16th Feb, Transfer learning for animal facial key point detection

Dr Marwa Mahmoud will present the paper Interspecies Knowledge Transfer for Facial Keypoint Detection published in CVPR 2017. Paper accessible here. Code and dataset accessible here.

22th Feb, Image quality assessment with CNNs

Aliaksei Mikhailiuk will present the paper The Unreasonable Effectiveness of Deep Features as a Perceptual Metric, published in CVPR 2018. Paper accessible here. Code and dataset accessible here.

17th April, Deep image prior for image enhancement

Dingcheng Yue will present the paper Deep Image Prior. Paper accessible here. Code and dataset accessible here.

24th April, Deep learning for point cloud estimation

Eva Agapaki will present the paper PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation. Paper accessible here. Code and project page accessible here.

8th May, Deep learning for image quality assessment

Nanyang Ye will present the paper Deep Learning of human visual sensitivity in image quality assessment, published at CVPR 2017. Paper accessible here. Code accessible here.

15th May, Data augmentation and empirical risk minimization

I will present the paper Mixup: Beyond empirical risk minimization, published at ICLR 2018. Paper accessible here. Code accessible here.

The "adas" (generations of women who have followed Ada Byron in her desire of breaking the mold, of creating and programming), prefer the structure "switch...case" over the structure "if...else", given its infinite possibilities. Programming the world for the “adas” implies creating the conditions for each one to be whatever one wants to be, even when this corresponds to something not referenced yet with a word in the world, or when it can not be imagined clearly; it implies creating the conditions that even allow us to change our mind.

Top