Dear Christos Chronis, Alexia Apostolakis, Simon Vellas
I hope this email finds you well
Hello my name is Aarav Maheshwari and I am second year undergraduate student at the department of Geology and Geophysics enrolled in its geophysics course with AI specialisation at Indian Institute of Technology Kharagpur. I would love to work on the thermal satellite ai framework project.
I have contributed in repos like tensorflow (worked on fixing kernel level issue and optimising it), hugging face transformer (integrating mimo model in transformers library) and scikit-learn (threading issue). I have a good amount of experience in training and integrating ml models so I would love to learn and make a completely new framework from scratch.
I had some questions regarding this project
Firstly I took look into Orion AI Lab repo and found Gaia, Thalia, TIRAuxCloud datasets — other than these do we have to integrate other standard thermal imaging datasets (like Cloud-SEN12 etc)?
I also saw televit1.0 pre-trained model can we integrate such specialised model except base models,(would love any suggestions for some specialised models like televit1.0 that can be integrated)?,Some of the base models I was thinking of integrating in this project were (U-net, YOLO, Segformer, Masked Auto Encoder, CLIP) — any other suggestions for base model that should be added?
Also for data augmentation aside from basic techniques is there any specific technique that is used in case of thermal image data that has to be implemented (any research papers pointing to any such techniques would be really helpful)?, I looked into some and have implemented there kernels (https://github.com/Aaraviitkgp/gsoc-gfoss-sample) I have implemented kernels for cpu only do we have to add cuda integration as well ?
Best Regards
Aarav Maheshwari
----
Λαμβάνετε αυτό το μήνυμα απο την λίστα: Λίστα αλληλογραφίας και συζητήσεων που απευθύνεται σε φοιτητές developers \& mentors έργων του Google Summer of Code - A discussion list for student developers and mentors of Google Summer of Code projects.,
https://lists.ellak.gr/gsoc-developers/listinfo.html
Μπορείτε να απεγγραφείτε από τη λίστα στέλνοντας κενό μήνυμα ηλ. ταχυδρομείου στη διεύθυνση <gsoc-developers+unsubscribe [ at ] ellak [ dot ] gr>.