WittyTourist is a web application that gives you a witty caption when you upload a photo with a San Francisco landmark. Using computer vision, the app identifies the landmark in your photo and returns a witty caption for that landmark.
You’ve taken the perfect photo and you’re ready post to social media, but now comes the hard part…coming up with a witty caption!
- Step 1 - Snap Pic
- Step 2 - Get Witty Caption
- Step 3 - Post to social media and watch your likes roll in like Karl on a foggy day in San Francisco.
The app is currently trained to detect these San Francisco landmarks (more coming!):
- The Golden Gate Bridge
- The Oakland Bay Bridge
- A cable car
- Lombard Street
- Alcatraz
- The Painted Ladies at Alamo Square
- The Palace of Fine Arts
- The sea lions at Pier 39
- The Transamerica Pyramid
- Muir Woods
- Ghirardelli Square
- Coit Tower
- Fisherman's Wharf sign
Blog Post: 9 Steps to Building a Deep Convolutional Neural Net in Excel for Normal Humans.
Cutting-Edge Face Recognition is Complicated. These Spreadsheets Make it Easier.
The backbone architecture of this convolutional neural net (CNN) is ResNet-50. The model was initially trained on ImageNet and using transfer learning, the weights were re-trained and fine-tuned to fit the landmark classes in this app. This is a multi-class image classification algorithm.
To learn how to build your own app, check out these resources:
- Video lesson from fast.ai - How to build an image classification model
- Model deployment - Instructions on how to deploy an app on Render