The problem SMLR solves

Online conferencing has played a major role in allowing educational institutions to conduct classes despite the pandemic. However it does possess drawbacks. Students face difficulties in paying attention for prolonged periods of time over a virtual setup, which ultimately leads to them missing out on learning important topics.

We aim to aid the students in this aspect, to ensure they do not miss out on anything.

SMLR (a smaller version of smaller xD ) is our solution to this problem. We provide a web based assistant that extracts the most important information from a session , saving time while ensuring students never miss out on the topics that are crucial.

How we add value:
-SMLR processes the videos, generating comprehensive summaries and notes of topics discussed in class, ensuring that students keep track of what's important. An addition to this is an automated integration with DropBox, allowing students to access the information whenever they want to.
-SMLR's key feature is an automated question generator, powered by a transformer Neural Network architecture, which uses the notes that it generates to suggest probable questions that can be asked in exams based on the content covered in that session, alongside with the ideal answers to the generated questions. We believe this feature is a great value addition, which ultimately provides a better learning experience.
-Ease of use. The user just has to upload the video to our service and SMLR takes care of the rest.
-Dashboard to organize the processed lectures.
-Push notifications to notify the user after processing

Challenges we ran into

1. The time constraint: Integrating multiple features into the application within the span of a day was quite challenging, leaving little time for testing it. However, it was also a good learning experience.
2. We faced a few issues integrating Video to text with the backend, had to rely on free service providers.
   3)T5 transformer can handle only 512 tokens max, so we had to iterate through the text and run it multiple times to get the full summary. We will also lose some valuable contexts when splitting