Summary of Hours and Justification

Achyut Anand

I worked for about 135 hours each semester, so 270 hours in total. Out of those 270 hours, roughly 40 hours were spent in meetings with my teammates and our project advisor Rashmi Jha. I spent 50 hours researching for the project that included going through numerous research papers, collecting and testing datasets for our use cases. About 25 hours were spent developing the semi-supervised road sign model and the GPS models. Another 25 hours were spent creating other machine learning models that we did not end up using in our final product. 30 hours were spent developing the front-end code of our demo. 10 hours were spent connecting the back-end code to the front-end for our demo. 40 hours were spent working on class assignments, 25 hours in the fall semester, and 15 in the spring. 10 hours were spent working on the presentation and demo videos for the Senior Design Expo. 20 hours were spent testing our final product. And lastly, 20 hours were spent trying to work with a tool that eventually did not work for our project. Those tools included Azure, Google Cloud Platform, Automotive Grade Linux, and IBM Maximo.

Zach Hammitt

I estimate that I spent around 130 hours each semester myself. Much of this was spent on bureaucratic and logistical things like meetings, research, and testing. Meetings accounted for about 40 total of those hours as there was roughly a meeting a week. Research and testing took quite a bit of the time as well. As I didn't have much of a clue what I was getting into at the beginning of all of this, I had to look up various AI algorithms and try to understand them and why we should use which one. Then there was also hunting down datasets which proved to be quite difficult on a budget of 0 dollars. On a less technical side, much research time was spent on looking into and trying all the various tools that exist out there, many of which didn't get used such as IBM Maximo, Automotive Grade Linux, and Microsoft Azure. Overall, I would say the research part took around 60 hours split rather evenly between the semesters. Designing the Unity Demo at the end took about 20 hours on its own to perfect. The assignments required for class like creating diagrams, test plans, and this very wiki, took maybe 40 hours total with much of this front loaded towards the first couple months of the project. Lastly, I would say I spent roughly 100 hours total actually working on our code and projects. The majority was spent on creating and then testing and improving the road sign and object detection modules. The road sign module got done around November and December while the Object Detection module was worked on all through January and February. Some time was also spent of course on the GPS module and helping to integrate the final product.

Mathew Lucy

My work totaled an average of 80 hours a semester doing work and 30-40 hours a semester doing meetings, to a total of about 220 hours in the year. The meetings included meetings with my full group, pair programming work on certain projects, and meetings to update and present to our advisor Dr. Rashmi Jha. For the first semester much of the time was spent doing research to understand how to best approach the project and what existing work we could build off of, totaling a little over half of the time spent in the first semester (40-50 hours). The second half of the first semester was spent working on finding datasets for each of our modules and working with the first dataset, the German Roadsign dataset. My part specifically was developing the unsupervised K-Nearest Neighbor System, which took around 30-40 hours. This took the longest of any module due to the fact that it was the first, and much of the code for later modules was reused from it. The second semester had a multitude of tasks, including creating the second module, a histogram of ordered gradients on the roadsign object detection KITTI dataset, totaling around 20 hours. Thirty hours were then spent researching how to compress models for the backend, backend development, and working to improve accuracy and speed for the models. Finally, another 10 hours were put into hooking up the model with the front end and performing testing to make sure the project would work for the demo. And additional 20 hours were spent doing class work such as cleaning up and updating the github, as well as preparation for the senior design expo and CEAS Showcase which we additionally participated in. Overall, my work was evenly split throughout the year between research, design, coding, testing, and presentation work.