Title: Distributed Cloud/Edge AI solution for video analysis
Department: HMS Labs
Size: 1-4 students
Scope: Computer vision, Artificial Intelligence
Description: There is an increasing interest in intelligent applications for industrial use cases. One area where smart AI-driven applications are of particular interest is for the collection of information from the production floor.
Video cameras are attractive as sensors to collect data from the factory floor because they are both cost efficient and do not require physical connection to the production machines. Traditionally the cameras are connected through wired Ethernet to a server which performs the image analysis. In order to increase flexibility there is a growing need to connect the cameras using wireless technology. This, however, poses a problem because the wireless connection are limited in throughput.
The goal of this thesis project is to develop a distributed computer vision solution which does pre-processing at the edge and more general purpose AI in the cloud. The project should consider typical industrial AI use cases such as "visual quality inspection", "counting number of produced units", "state of production machine" and "unexploded dynamite during mine blasting". Please check this video for a description of the "finding holes in a fence" use case: https://youtu.be/3u_DUKTR014
It is a goal of the thesis work to create a computer vision edge device with the following characteristics:
- The edge device should be generic and cover all common use cases.
- The edge device should be under the control by an application specific cloud AI solution.
- The edge/cloud communication should be optimized to minimize data transfer.
- The edge device should be optimized for battery operation and minimum energy consumption.
The the edge device should be integrated with a cloud AI solution in order to demonstrate one or two typical use cases.
The project and scope can be adjusted to the size of the group and the participants.
Prerequisites:
Bachelor or master program within computer science or similar. Experience in a programming environment suitable for AI development like Python or Matlab, knowledge in artificial intelligence and computer vision, and experience in C/C++, make, CMake is an advantage.
Contact Person:
Jens Jakobsen
+46 7215 969 25
JeJa@hms.se