Designing Compact Deep Neural Networks for Multimedia Analysis

12th September 2019
Designing Compact Deep Neural Networks for Multimedia Analysis

Proposed by Aakanksha Rana and Cagri Ozcinar Email: ranaa at or ozcinarc at

With the emergence of the Internet of Things (IoT) and powerful deep-learning-based multimedia analysis algorithms, compact deep neural network models are crucial for implementing them on small embedded devices. Existing models developed for multimedia analysis are computationally expensive and memory intensive, hindering their deployment in small embedded devices with low memory resources. Typical model compression solutions are network pruning and weight quantization techniques. In this project, we will investigate algorithms developed for model compression and develop a novel technique to reduce the size of the network for multimedia analysis.



Basic understanding of Deep-learning,

Strong Python programming skills with knowledge of Pytorch/TensorFlow/Keras tools.

The ideal candidate should have an interest in compression and deep learning algorithms in general and must have the ability to learn new tools and knowledge.