The topic of this course is the detection of humans (and other objects) in images.
Object detection is one of the most challenging tasks in the field of computer vision and machine learning. The difficulty is because many objects have complex appearances; for instance, humans often adopt varying poses, and have different sizes.
The goal of this project is the detection of object instances in images using local features and supervised learning methods. The students will implement a detector for humans which performs localization by specifying a tight bounding box around each instance.
During this project you will learn
- the basics of deep neural networks
- an approach to object detection detection based on deep neural networks
- to use the TensorFlow machine learning framework
- how to objetively evaluate the performance of an object detector
This course is divided into two phases:
- Assignments (handed out every week) will introduce students to programming with Python, step-by-step build the detection pipeline and provide first hands-on experience in image processing and machine learning.
- Student groups will work on a project in weekly sessions
In the end, each team will present the results of their implementation to the other teams.
- Finding People in Images and Videos, Navneet Dalal. PhD Thesis. Institut National Polytechnique de Grenoble / INRIA Grenoble , Grenoble, July 2006.
- SSD: Single Shot MultiBox Detector, W. Liu, D. Anguelov, D. Erhan, C. Szegedy, S. Reed, C. Fu, A. Berg, ECCV 2016