Deep Learning for Computer Vision
Submitted by Anand Chandrasekaran (@anandchandrasekaran) on Wednesday, 8 June 2016
One of the fields that have benefited the most from the rise of Deep Learning has been Computer Vision. The goal of this workshop is to have participants go from the basics to tackling a problem that might solve a real world problem.
Some of the theory covered will be:
- From neurons to networks, a full overview of the nuts and bolts.
- Types of networks, from RBMs to CNNs to RNNs.
- How they are used in the world of CV. A discussion of what works and what doesn’t.
- The fundamentals of training Deep Learning networks.
- On existing frameworks and using GPUs.
A relatively large number of potential projects will be available for the workshop, and the participants will have access to AWS ec2 GPU instances for training and testing.
Some Potential workshop projects:
- Toy problems exploring the different architectures and relative merits.
- Real world classification problems solved with different architectures.
- Exploring novel and hybrid architectures (if time permits).
The participants need a laptop with:
- Python 2.7
- Numpy 1.8+
- OpenCV 2.4.9+
For the heavy lifting, AWS instances with all the dependencies and the datasets will be setup for participants to focus on their python scripts, and not worry about installations.
For the intrepid, with their own GPU laptops we can provide instructions for local installations of any framework used, as well as the data sets.
Dr. Anand Chandrasekaran is a founder and the CTO of Mad Street Den, an AI company specializing in computer vision. In addition to an academic background in the fields of neuroscience and neuromorphic engineering, he has been a member of teams working on DARPA projects in cognition and vision.