Thu, Apr 5, 2018
A couple of weeks ago we received an invitation from Adrian Rosebrock to give a workshop at the first edition of the PyImageConf in San Francisco.
Our first reaction was of excitement.
We were already aware of the great speakers that were featured (François Chollet, Davis King, Adam Geitgey, Adrian Rosebrock himself, among others), and we really liked the format of the event: practical & intimate.
Moreover, the conference is centered on three fields we're pretty familiar with: Deep Learning, Computer Vision & Python.
After the initial excitement, we accepted the invitation from Adrian and started thinking about the topic for the workshop.
We ended up deciding for a hands-on workshop on how to use Deep Learning for Object Detection.
We selected this topic since:
We've been tackling quite a few projects on this area lately.
We've been sharing our experiences in our blog posts and giving talks at different conferences.
Most importantly, we have recently launched an open-source Python toolkit, Luminoth, which aims to facilitate training, monitoring and deployment of Deep Learning models for Object Detection.
Basically, the idea is to divide the workshop into 2 segments: some theory and practice.
Theory: we'll start with a brief historic review, we will discuss the most recent DL models from the relevant research papers. We will go into Faster R-CNN in-depth, a model that has state of the art performance, by learning the architecture and going over an actual TensorFlow implementation in Luminoth, an open-source Python toolkit for Computer Vision.
Practice: with all the knowledge acquired, the audience is going to learn how to train and evaluate DL models for detecting cars in images and videos, and for detecting & classifying cars, buses, bikes and pedestrians.
The workshop will be given by Alan Descoins (CTO) and Agustín Azzinnari (Research Engineer).
For the people attending the workshop, we recommend the following:
Familiarity with Python (must), and Jupyter Notebooks (optional).
Basic knowledge of Machine Learning.
Helpful to have:
Basics of Deep Learning and Convolutional Neural Networks (CNN).
Familiarity with TensorFlow.
For those of you interested, the workshop will be hosted on Tuesday, August 28th at the Hyatt Regency and will last for the whole day. You can learn more about the PyImageConf, the talks, and buy your tickets at pyimageconf.com.
All in all, this will be an exciting event for us. Not only because we'll be sharing our experiences building object detection systems with Deep Learning, but also because we'll be enjoying several talks from very experienced people in the fields of Computer Vision, Deep Learning & Python.
We're very much looking forward to it. 😊
Along with Microsoft, Clusterone & Colorhythm, we're also thrilled to support the PyImageConf through sponsorship. Hopefully, this edition is the first of many more to come!