Warning: count(): Parameter must be an array or an object that implements Countable in /srv/users/serverpilot/apps/eken/public/th_main/class/Single/SinglePost.php on line 1081
The description here is based on the course “AT 70.20: Machine Vision for Robotics and HCI” taught in AIT by Professor Mathhew Dailey in Computer Science and Information Management department. I was much impressed by the project topics so posting them here.
Machine vision is concerned with the image processing, geometry, and statistical interference tools necessary for extracting useful information about the world from two-dimensional images. After decades of research, although the most advanced machine vision systems still pale in comparison to the visual systems of the simplest mammals, there have been some success stories.
- Augmented Reality: understand the scene and embed computer graphics in it.
- 3D image processing: given a CT scan of a jaw and a tooth number, recognize the structure, segment the teeth, get the internal structure, and build a surface model of other teeth and voxel model of the tooth of interest.
- Intelligent video surveillance
- Smooth tracking of an arbitrary pre-specified object in a video sequence.
- Obstacle detection for a mobile robot
- Vehicle detection and tracking for intelligent transportation systems. Using a camera mounted on a vehicle, detect and track other nearby vehicles.
- 3D reconstruction of a known object from a partial view in a video with occlusions. The object is a pineapple.. The idea is we see a pineapple in a few video frames and have to determine the 3D position, orientation, and scale from those frames.
- 3D detection of objects from skeleton. The idea is we have a factory automation situation in which a bin is piled full of objects with known geometry, and we want to find the topmost part and pull it out (for integration with an assembly line, for example). Here we’ll begin with a 3D model of the part and use robust estimation to fit the model to the observed image.
These projects can be carried out focusing on 3D reconstruction, machine learning in vision, and sequential state estimation (tracking). Matlab/Octave and the OpenCV library can be used for the implementation.