For a couple of semesters now, I have helping a team under the mentorship of prof. Vineeth Balasubramanian and prof. C.V. Jawahar for a Workshop submission for CVPR 2021. The workshop corresponds to an autonomous navigation challenge AUTONUE 2021. Any readers interested in participating may follow the instructions listed here, with a total prize pool of 5000 USD sponsored by Intel.
I’ll briefly touch upon my experience and the work I did…
- I was majorly involved in devising the problem statement for the challenge.
- This involved researching the characteristics of multiple source datasets available and choosing the best combination for the challenge.
- I also helped implement the baselines for the various subsettings in the challenge, this was the major part of my work.
- Moreover, based upon the results of the baselines, we selectively prepared the datasets and all the convenience scripts that would be potentially useful to participators.
- Finally, now that the challenge is live, we clear any doubts or questions that come up by the participators on the challenge forums.
I worked on the Domain-Adaptation front (4 sub-challenges out of 5) of the challenge and this was my first formal research experience in the academia. All in all, got to know about the process of iterating research and discovering academic problems while coordinating with multiple academics at a time.
When I began this project, I was unsure about how to get about with the specifics of the work but the aforementioned mentors were considerably patient and that eased of the process. Now, I’m still working with prof. Vineeth, exploring the field along the same lines and looking forward to whatever that leads to.