This video contains proprietary information and cannot be shared publicly at this time.
Figure 1
Computer Science and Engineering
Team 33
Team Members |
Faculty Advisor |
Ryan Akhundzadeh |
Hanna Aknouche-Martinsson Sponsor General Dynamics: Electric Boat |
sponsored by
The goal of this project was to utilize Artificial Intelligence, specifically a Convolutional Neural Network (CNN), to identify a specific symbol that Electric Boat utilizes during the building process. Ultimately, the final program would be able to locate and return the center pixel of the symbol in a given image. The CNN was designed utilizing both Python and Keras. Supervised training of the CNN was conducted using digital recreations of the symbol that were produced by a program that was able to manipulate the homography of a given image. The layers of the CNN were augmented throughout the design process to achieve greater accuracy and reduce over fitting. Once the CNN was trained, further work was done to allow the CNN to identify specific features of symbol, which in turn allowed the image to identify and return the center pixel.