Inspiration
Tribal nature in sports audience.
What it does
Detects crowds. Scope: Could add filters like Snap filters.
How I built it
We iteratively kept making changes to
- Training Data distribution
- Data Augmentation
- Model Layers
Challenges I ran into
Modifying the size of each layer in the ConvNet.
Accomplishments that I'm proud of
We thought through the problem!
What I learned
Learned a whole lot about tweaking conv nets.
What's next for Crowd Recognition & Enhancement w/ Data
Scope:
- add enhancements like filters
- enable sentiment prediction
Built With
- keras
- python
Log in or sign up for Devpost to join the conversation.