About This Topic
Develop solutions for the recognition of defined visual objects in video data streams based on deep learning and a commercial video analysis cloud (Google Precision Cloud and IBM Bluemix preferred). Requires the provision of examples, neural network design, implementation in the cloud environment and extended testing of the performance. Requires a good understanding of state-of-the-art methods for large scale network-based video analysis (e.g. Tensor Flow, the map reduce algorithm, Hadoop etc.). This topic is also suitable for groups of 2-3 students.