LectureGenie

Tired of lectures that go in one ear and out the other? LectureGenie employs AI to make learning interactive, turning passive watching into active engagement for students to truly own their education.

Comment

Inspiration

Our inspiration comes from the common experience of getting distracted while watching recorded lecture videos. We wanted to create a tool that helps viewers stay focused, actively engage with the material, and ultimately gain a deeper understanding of the content.

What it does

LectureGenie is a software program specifically designed to enhance the learning experience with recorded lectures. It takes a lecture video as input and generates several outputs to aid understanding. It works with both silent videos and those with audio:

  • Quiz questions: As the user watches the video, they can ask for a quiz question (Multiple Choice or T/F) to be generated based on the nearby frames and recent audio. By answering questions while watching, comprehension and attention are greatly improved.
  • Notes: LectureGenie automatically summarizes the key concepts and information presented in the video, creating a concise and easy-to-reference set of notes.
  • Flashcards: Based on the identified key points, the program can also generate flashcards, aiding in memorization and efficient review.

How we built it

Written in Python, using Streamlit for the front end. The main server is hosted on an AWS EC2 instance. Several API calls are made to both Gemni and Gemin-Vision to process the information in the video. An AWS transcript generator endpoint is used to create transcripts.

Challenges we ran into

Keyframing -> Figuring out how to pick only the important frames to process without discarding too many.
Transcribing -> How do you get a transcription and process it from a video?
Frame Descriptions -> How to use multiprocessing and set up our own API in order to make many concurrent calls to Gemini?

Accomplishments that we're proud of

  • Building a tool we feel we would actually use
  • Making a visually appealing website
  • Deploying to AWS
  • Integrating with Gemini and parsing its output
  • Successfully utilizing Amazon Transcribe

What we learned

  • About the complexities involved in processing and analyzing video content for educational purposes.
  • The importance of user experience in designing effective learning tools.
  • The potential of technology to enhance the traditional lecture format and support self-paced learning.

What's next for LectureGenie

  • Improving the accuracy and effectiveness of content analysis and knowledge extraction.
  • Incorporating more types of questions, such as matching and fill-in-the-blank.
  • Allowing users to customize the type and difficulty of generated quizzes and flashcards.
Share this project:

Updates