Content based web image retrieval
Synopsis: With the prevalence of images online, finding the related or intended image becomes important for users. Users used to search images only by keywords or context. However, image carries more information. If user has one sample image, how to query the Internet to find similar images is an interesting problem. Google has published their image-based search engine. We want to accomplish similar effects.
In this project you will:
1. Find related work on content based image retrieval
2. Test existing image search engines.
3. Propose and implement a simple algorithm to retrieval similar images from Internet (like flickr) based on keywords and sample images.
 Liu, Ying, et al. “A survey of content-based image retrieval with high-level semantics.” Pattern Recognition 40.1 (2007): 262-282.
 Perronnin, Florent, et al. “Large-scale image retrieval with compressed fisher vectors.” Computer Vision and Pattern Recognition (CVPR), 2010 IEEE Conference on. IEEE, 2010.
Deliverables: In the end of the semester, the student should provide a written report on the work done as well as the implemented algorithm that can perform image retrieval based on keywords and sample images.
Prerequisites: Basic knowledge of image processing and computer vision, programming with Matlab or C/C++ using OpenCV.
Type of work: 50% research and 50% implementation
Level: MS, semester project, Computer Science or Communication Systems.
Supervisor: Bin Jin (firstname.lastname@example.org).
Evaluation of user experiences with wearable augmented reality in fire fighting operations
Firefighters use thermal cameras to better orient and scan for hotspots and casualties. In an ongoing project, an augmented reality application has been developed that would allow a fire fighter to see a thermal heat map on a wearable see-through display.
The student will:
– Prepare, conduct and evaluate usability tests with augmented reality prototypes. The tests will be performed on different locations, including fire fighting testing facilities outside of EPFL.
– The main goal of the tests is to quantify if AR improves the efficiency of fire fighters on typical tasks, for example inspired by the tests in .
– Perform integrated test of localization, communication, real-time video links and AR, using components developed in previous projects.
– Where needed, program missing missing bits of functionality, or add features that add to the overal usability of the system.
 J. Boyd, “Using Hands-Free Thermal Imaging Cameras,” Fire Eng., vol. 160, no. May, 2007.
Number of students: 1
The student should provide in the end of the semester a written report related to the work done. The whole soft- and hardware implementation used for generating the results in the report should also be provided.
Prerequisites: Programming experience in Android/Java. Affinity with fire fighting.
Type of work: 30% Research, 30 Tests, 40% implementation
Level: MS, Semester project
Supervisors: Adrien Birbaumer, Martijn Bosch
Detect inserted video
Description: This is a video matching problem. During the course of video production, a portion (a set of frames) of one video may be inserted into another video. The goal is to find this set of inserted frames giving the original video and the one that contains the inserted frames.
Note: This is an RTS (radio television Swiss) project.
1. Set up a video decoding scheme (ffmpeg or such)
2. Code-up a few standard algorithms for matching frames
An executable (command line or GUI) that can take videos as input and indicate matches
Number of students: 1
– knowledge of image processing, image retrieval etc.
– coding skill in C / C++
Level: BS / MS semester project, Computer Science or Communication Systems
Type of work: 30% research and 70% implementation
Supervisor: Radhakrishna Achanta (email@example.com)