Deep Learning in Computer Vision

Topics in Computer Vision (CSC2523):

Deep Learning in Computer Vision

Winter 2016

In recent years, Deep Learning has become a dominant Machine Learning tool for a wide variety of domains. One of its biggest successes has been in Computer Vision where the performance in problems such object and action recognition has been improved dramatically. In this course, we will be reading up on various Computer Vision problems, the state-of-the-art techniques involving different neural architectures and brainstorming about promising new directions.

Please sign up here in the beginning of class.

Course overview

This class is a graduate seminar course in computer vision. The class will cover a diverse set of topics in Computer Vision and various Neural Network architectures. It will be an interactive course where we will discuss interesting topics on demand and latest research buzz. The goal of the class is to learn about different domains of vision, understand, identify and analyze the main challenges, what works and what doesn't, as well as to identify interesting new directions for future research.

Prerequisites: Courses in computer vision and/or machine learning (e.g., CSC320, CSC420, CSC411) are highly recommended (otherwise you will need some additional reading), and basic programming skills are required for projects.

 back to top

Course Information

• Time and Location

  Winter 2016

  
  

  Day: Tuesday
  Time: 9am-11am
  Room: ES B149 (Earth Science Building at 5 Bancroft Avenue)
  

  Instructor

  Sanja Fidler

  
  

  Email: fidler@cs dot toronto dot edu
  Homepage:  http://www.cs.toronto.edu/~fidler
  Office hours: by appointment (send email)

When emailing me, please put CSC2523 in the subject line.

Forum

This class uses piazza. On this webpage, we will post announcements and assignments. The students will also be able to postquestions and discussions in a forum style manner, either to their instructors or to their peers.

 back to top

Invited Speakers

We will have an invited speaker for this course:

• Raquel Urtasun
  Assistant Professor, University of Toronto
  Talk title:  Deep Structured Models

as well as several invited lectures / tutorials:

• Yuri Burda, Postdoctoral Fellow, University of Toronto:    Lecture on Variational Autoencoders
• Ryan Kiros, PhD student, University of Toronto:    Lecture on Recurrent Neural Networks and Neural Language Models
• Jimmy Ba, PhD student, University of Toronto:    Lecture on Neural Programming
• Yukun Zhu, Msc student, University of Toronto:    Lecture on Convolutional Neural Networks
• Elman Mansimov, Research Assistant, University of Toronto:    Lecture on Image Generation with Neural Networks
• Emilio Parisotto, Msc student, University of Toronto:    Lecture on Deep Reinforcement Learning
• Renjie Liao, PhD student, University of Toronto:    Lecture on Highway and Residual Networks
• Urban Jezernik, PhD student, University of Ljubljana:    Lecture on Music Generation

Requirements

Each student will need to write two paper reviews each week, present once or twice in class (depending on enrollment), participate in class discussions, and complete a project (done individually or in pairs).

Grading

The final grade will consist of the following
Participation (attendance, participation in discussions, reviews)	15%
Presentation (presentation of papers in class)	25%
Project (proposal, final report)	60%

Detailed Requirements   (click to Expand / Collapse)

 back to top

Syllabus

The first class will present a short overview of neural network architectures, however, the details will be covered when reading on particular topics. Readings will touch on a diverse set of topics in Computer Vision. The course will be interactive -- we will add interesting topics on demand and latest research buzz.

Tentative Syllabus    (click to Expand / Collapse)

 back to top

Schedule (tentative)

Schedule

Date	Topic		Reading / Material		Speaker	Slides
Jan 12	Admin & Introduction(s)				Sanja Fidler	admin
Convolutional Neural Networks
Jan 19	Convolutional Neural Nets(tutorial)		Resources: Stanford's cs231 class, VGG's Practical CNNTutorial Code: CNN Tutorial for TensorFlow, Tutorial for caffe, CNNTutorial for Theano		Yukun Zhu (invited)	[pdf] [code]
	Image Segmentation		Semantic Image Segmentation with Deep Convolutional Nets and Fully Connected CRFs   [PDF] [code] L-C. Chen, G. Papandreou, I. Kokkinos, K. Murphy, A. L Yuille		Shenlong Wang	[pdf] [code]
Jan 26	Very Deep Networks		Highway Networks  [PDF] [code] Rupesh Kumar Srivastava, Klaus Greff, Jurgen Schmidhuber Deep Residual Learning for Image Recognition  [PDF] Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun		Renjie Liao (invited)	[pdf]
	Object Detection		Rich feature hierarchies for accurate object detection and semantic segmentation   [PDF] [code] Ross Girshick, Jeff Donahue, Trevor Darrell, Jitendra Malik Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks   [PDF] [code (Matlab)] [code (Python)] Shaoqing Ren, Kaiming He, Ross Girshick, Jian Sun		Kaustav Kundu	[pdf]
Feb 2	Stereo Siamese Networks		Stereo Matching by Training a Convolutional Neural Network to Compare Image Patches  [PDF] [code] Jure Žbontar, Yann LeCun Learning to Compare Image Patches via Convolutional Neural Networks  [PDF] [code] Sergey Zagoruyko, Nikos Komodakis		Wenjie Luo	[pdf]
	Depth from Single Image		Designing Deep Networks for Surface Normal Estimation   [PDF] Xiaolong Wang, David Fouhey, Abhinav Gupta		Mian Wei	[pptx]  [pdf]
Feb 9	Image Generation		Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks   [PDF] Alec Radford, Luke Metz, Soumith Chintala Generating Images from Captions with Attention   [PDF] Elman Mansimov, Emilio Parisotto, Jimmy Lei Ba, Ruslan Salakhutdinov		Elman Mansimov (invited)	[pdf]
	Domain Adaptation, Zero-shot Learning		Simultaneous Deep Transfer Across Domains and Tasks   [PDF] Eric Tzeng, Judy Hoffman, Trevor Darrell Predicting Deep Zero-Shot Convolutional Neural Networks using Textual Descriptions   [PDF] Jimmy Ba, Kevin Swersky, Sanja Fidler, Ruslan Salakhutdinov		Lluis Castrejon	[pdf]
Recurrent Neural Networks
Feb 23	RNNs and Neural Language Models		Unifying Visual-Semantic Embeddings with Multimodal Neural Language Models   [PDF] [code] Ryan Kiros, Ruslan Salakhutdinov, Richard Zemel Skip-Thought Vectors   [PDF] [code] Ryan Kiros, Yukun Zhu, Ruslan Salakhutdinov, Richard S. Zemel, Antonio Torralba, Raquel Urtasun, Sanja Fidler		Jamie Kiros (invited)
Mar 1	Modeling Words		Efficient Estimation of Word Representations in Vector Space  [PDF] [code] Tomas Mikolov, Kai Chen, Greg Corrado, Jeffrey Dean		Eleni Triantafillou	[pdf]
	Describing Videos		Sequence to Sequence -- Video to Text   [PDF] Subhashini Venugopalan, Marcus Rohrbach, Jeff Donahue, Raymond Mooney, Trevor Darrell, Kate Saenko		Erin Grant	[pdf]
	Image-based QA		Ask Your Neurons: A Neural-based Approach to Answering Questions about Images   [PDF] Mateusz Malinowski, Marcus Rohrbach, Mario Fritz		Yunpeng Li	[pdf]
Mar 8	Variational Autoencoders		Auto-Encoding Variational Bayes   [PDF] Diederik P Kingma, Max Welling Tutorial: Bayesian Reasoning and Deep Learning   [PDF] Shakir Mohamed		Yura Burda (invited)	[pdf]
	Text-based QA		End-To-End Memory Networks   [PDF] Sainbayar Sukhbaatar, Arthur Szlam, Jason Weston, Rob Fergus		Marina Samuel	[pdf]
	Neural Reasoning		Recursive Neural Networks Can Learn Logical Semantics   [PDF] Samuel R. Bowman, Christopher Potts, Christopher D. Manning		Rodrigo Toro Icarte	[pdf]
Mar 15	Neural Programming		Neural GPUs Learn Algorithms   [PDF] Lukasz Kaiser, Ilya Sutskever Neural Programmer-Interpreters   [PDF] Scott Reed, Nando de Freitas Neural Programmer: Inducing Latent Programs with Gradient Descent   [PDF] Arvind Neelakantan, Quoc V. Le, Ilya Sutskever		Jimmy Ba (invited)
	Conversation Models		A Neural Conversational Model   [PDF] Oriol Vinyals, Quoc Le		Caner Berkay Antmen	[pdf]
	Sentiment Analysis		Recursive Deep Models for Semantic Compositionality Over a Sentiment Treebank   [PDF] Richard Socher, Alex Perelygin, Jean Y. Wu, Jason Chuang, Christopher D. Manning, Andrew Y. Ng and Christopher Potts		Zhicong Lu	[pdf]
Mar 22	Video Representations		Unsupervised Learning of Video Representations using LSTMs  [PDF] Nitish Srivastava, Elman Mansimov, Ruslan Salakhutdinov		Kamyar Ghasemipour	[pdf]
	CNN Visualization		Explaining and Harnessing Adversarial Examples   [PDF] Ian J. Goodfellow, Jonathon Shlens, Christian Szegedy		Neill Patterson	[pdf]
Mar 29	Direction Following (Robotics)		Listen, Attend, and Walk: Neural Mapping of Navigational Instructions to Action Sequences   [PDF] Hongyuan Mei, Mohit Bansal, Matthew R. Walter		Alan Yusheng Wu	[pdf]
	Visual Attention		Recurrent Models of Visual Attention   [PDF] Volodymyr Mnih, Nicolas Heess, Alex Graves, Koray Kavukcuoglu		Matthew Shepherd	[pdf]
	Music		A First Look at Music Composition using LSTM Recurrent Neural Networks   [PDF] Douglas Eck, Jurgen Schmidhuber Deep Karaoke: Extracting Vocals from Musical Mixtures Using a Convolutional Deep Neural Network   [PDF] Andrew J.R. Simpson, Gerard Roma, Mark D. Plumbley		Charu Jaiswal	[pdf]
	Music generation		Overview of music generation		Urban Jezernik (invited)
	Pose and Attributes		PANDA: Pose Aligned Networks for Deep Attribute Modeling  [PDF] Ning Zhang, Manohar Paluri, Marc'Aurelio Ranzato, Trevor Darrell, Lubomir Bourdev		Sidharth Sahdev	[pptx]
	Image Style		A Neural Algorithm of Artistic Style   [PDF]  [code] Leon A. Gatys, Alexander S. Ecker, Matthias Bethge		Nancy Iskander	[pdf]
Apr 5	Human gaze		Where Are They Looking?   [PDF] Adria Recasens, Aditya Khosla, Carl Vondrick, Antonio Torralba		Abraham Escalante	[pdf]
	Instance Segmentation		Monocular Object Instance Segmentation and Depth Ordering with CNNs   [PDF] Ziyu Zhang, Alex Schwing, Sanja Fidler, Raquel Urtasun Instance-Level Segmentation with Deep Densely Connected MRFs  [PDF] Ziyu Zhang, Sanja Fidler, Raquel Urtasun		Min Bai	[pdf]
	Scene Understanding		Attend, Infer, Repeat: Fast Scene Understanding with Generative Models   [PDF] S. M. Ali Eslami, Nicolas Heess, Theophane Weber, Yuval Tassa, Koray Kavukcuoglu, Geoffrey E. Hinton		Namdar Homayounfar	[pdf]
	Reinforcement Learning		Playing Atari with Deep Reinforcement Learning   [PDF] Volodymyr Mnih, Koray Kavukcuoglu, David Silver, Alex Graves, Ioannis Antonoglou, Daan Wierstra, Martin Riedmiller		Jonathan Chung	[pdf]
	Medical Imaging		Classifying and Segmenting Microscopy Images Using Convolutional Multiple Instance Learning   [PDF] Oren Z. Kraus, Lei Jimmy Ba, Brendan Frey		Alex Lu	[pptx]
	Humor		We Are Humor Beings: Understanding and Predicting Visual Humor  [PDF] Arjun Chandrasekaran, Ashwin K Vijayakumar, Stanislaw Antol, Mohit Bansal, Dhruv Batra, C. Lawrence Zitnick, Devi Parikh		Shuai Wang	[pdf]

 back to top

Resources

Tutorials, related courses:

•   Introduction to Neural Networks, CSC321 course at University of Toronto
•   Course on Convolutional Neural Networks, CS231n course at Stanford University
•   Course on Probabilistic Graphical Models, CSC412 course at University of Toronto, advanced machine learning course

Software:

•   Caffe: Deep learning for image classification
•   Tensorflow: Open Source Software Library for Machine Intelligence (good software for deep learning)
•   Theano: Deep learning library
•   mxnet: Deep Learning library
•   Torch: Scientific computing framework with wide support for machine learning algorithms
•   LIBSVM: A Library for Support Vector Machines (Matlab, Python)
•   scikit: Machine learning in Python

Popular datasets:

•   ImageNet: Large-scale object dataset
•   Microsoft Coco: Large-scale image recognition, segmentation, and captioning dataset
•   Mnist: handwritten digits
•   PASCAL VOC: Object recognition dataset
•   KITTI: Autonomous driving dataset
•   NYUv2: Indoor RGB-D dataset
•   LSUN: Large-scale Scene Understanding challenge
•   VQA: Visual question answering dataset
•   Madlibs: Visual Madlibs (question answering)
•   Flickr30K: Image captioning dataset
•   Flickr30K Entities: Flick30K with phrase-to-region correspondences
•   MovieDescription: a dataset for automatic description of movie clips
•   Action datasets: a list of action recognition datasets
•   MPI Sintel Dataset: optical flow dataset
•   BookCorpus: a corpus of 11,000 books

Online demos:

•   Lots of cool Toronto Deep Learning Demos: image classification and captioning demos
•   Lots of cool demos for ConvNets by Andrej Karpathy
•   Reinforcement Learning with Neural Nets (read paper for more info)
•   Places: scene classification with neural nets
•   CRF as RNN: Semantic Image Segmentation
•   drawNet: visualization of ConvNet activations
•   Visualization of ConvNets for digit classification
•   AI-painter: modify your photo in a certain style (eg, Van Gogh); uses neural nets as explained in this paper

Main conferences:

•   NIPS (Neural Information Processing Systems)
•   ICML (International Conference on Machine Learning)
•   ICLR (International Conference on Learning Representations)
•   AISTATS (International Conference on Artificial Intelligence and Statistics)
•   CVPR (IEEE Conference on Computer Vision and Pattern Recognition)
•   ICCV (International Conference on Computer Vision)
•   ECCV (European Conference on Computer Vision)
•   ACL (Association for Computational Linguistics)
•   EMNLP (Conference on Empirical Methods in Natural Language Processing)