Inspiration
Sometimes it takes a lot of time to find good Instagram hashtags for your photo.
What it does
This app helps quickly find suitable hashtags for your picture.
How I built it
I build it with Android, TensorFlow Lite, MobileNet and Firebase ML vision image label model. I combine both predictions from MobileNet and Firebase ML model.
Challenges I ran into
ML models usually provide categories for images, but I wanted to provide popular hashtags. Thus, I had to collect on Instagram and different other resources, hashtags which are usually associated with the given category. For example, with category 'coffee', usually hashtags #love, #coffeetime, #cafe, #instagood,#coffeelover,#photography, #food, #coffeeshop are mentioned.
Accomplishments that I'm proud of
I collected a big database of hashtags which are mentioned in Instagram for different categories. My app is completely offline. It uses on device ML models.
What I learned
How to use TensorFlow and Firebase ML models on mobile devices.
What's next for EasyHash
Create better ML model for EasyHash. Provide for user on-device training the ML model, such that the recommendation of the hashtags will be customized for the user. Add suggestion of tags based on geolocation.
Built With
- android
- firebase-ml-vision
- tensorflow
Log in or sign up for Devpost to join the conversation.