MEng Projects

The VIA research group typically offers a number of MEng project opportunities each year. Projects may be done by one or several students. Most projects involve developing algorithms for computer vision and require programming with the VisionX and deep learning software systeasm. Therefore, there is usually a requirement that students starting in the Fall semester also take ECE5470 at the same time to gain background knowledge in computer vision and familiarity with the programming tools.

The VIA MEng project program is in two parts with the second conditional on the first. For the first semester the student will plan the project and conduct the background research. At the end of the semester an initial report is required; the continuation of the project to the second semester will be conditional on a satisfactory report submitted at the conclusion of the first semester. If time permits the work on the project research may also be started in the Fall semester. The main activity for the second semester is to conduct the actual research and write up a final report.

Project Topics for Fall 2019

The general project theme of the VIA group is to implement fully automated computer algorithms to identify clinically relevant diseases and issues in the chest region from 3D low-dose CT images. New for 2019 there will be two general robotics projects. All projects involve exploring solutions in computer vision including deep learning methods. Projects often involve groups of several students. The following projects are being offered for 2019:

  1. Analysis of pulmonary nodules in low-dose CT images
  2. Heart Health analysis (coronary artery calcium score) from CT images
  3. Experimental mobile vision robot that can do real-time image analysis and navigation: Navigation and Image Analysis
  4. Implementation of an advanced javascript image viewer with annotations for web browsers. Knowledge of Javascript and Ajax needed for this project.

Required Skills

Self starting person, motivated, interest in computer vision and machine learning; will need to take ECE 5470 “Computer Vision” in the Fall to gain experience in image analysis tools, Linux/UNIX, C programming and python.

Project Course Requirements

  1. Project groups will schedule a one hour weekly meeting time with A. P. Reeves for each semester. In general meetings will not be every week and typically average about eight a semester.
  2. Each semester each project group will be required to submit a draft report in mid-semester and a full report at the end of the semester.
  3. Project report deadlines:
    Fall semester draft report October 18
    Fall semester interim report December 12
    Spring semester draft report March 5
    Spring semester final report May 6

Students who are doing a two semester project will be expected to submit an Interim Project Report at the end of the first semester. Note, this report is more detailed than the requirements for the MEng program.

All students doing an MEng project will be required to submit an individual Project Report at the end of the project. Use the class project template for your report. Also review the project draft guidelines. In some cases a group may submit a single report if the indiviual effort of each group member is clearly identified in the report.

Students interested in doing an MEng project should send an email with their resume and a brief description of their interest to A. P. Reeves. The title of the email should include the words "MEng Project".

The Project Computer Server

The project computer server is rimmer.via.cornell.edu. All students doing independent projects have accounts on this server for project programming and testing. The environment on rimmer is similar to that for the ECE 5470 platforms: ecelinux-10 and the personal VM. Project image datasets are hosted on this server.

For unix access to the server use the following command from a suitable ssh client: 

     ssh -Y ‹NETID›@rimmer.via.cornell.edu

and log in with your Cornell netid password. There is also web-based access to the image database for convenient browsing and to facilitate image annotation. To access the web server go to: https://rimmer.via.cornell.edu/cgi-bin/datac/clogon.cgi?prog=X. Click on "Request Access", then enter your Cornell NETID credentials.

There is no regular scheduled backup for this server. Make sure that important files have backup copies stored elsewhere.