Inspiration
Since we all love dogs, we wanted to create a project that would be fun and useful for identifying and familiarizing ourselves with various dog breeds.
What it does
The user will upload a photograph of a dog to the website, and then the website will output the dog breed along with a generic photograph of the specified breed.
How we built it
The web application has a backend of Node.js and frontend of handlebars. It is run on TJHSST's Director under this link: https://user.tjhsst.edu/2021cli/home? For the code in charge of processing and organizing images, creating augmented images, and training the model, we modified code from this site: http://blog.stratospark.com/deep-learning-applied-food-classification-deep-learning-keras.html
Challenges we ran into
The computer ran out of storage to store all the images of the dog breeds, so we had to delete many of the images. Additionally, TJHSST Director is slow and often gave us bad gateways.
Accomplishments that we're proud of
We are proud of 100% accurately identifying dog breeds in a way that is constructive and helpful to anyone who wants to use our site.
What we learned
Cathryn's computer does not have much storage or enough computing power to do this project in a large scale.
What's next for Dog Breed Identifier
Add more dog breeds into the mix!
Built With
- css
- html
- javascript
- python
- tensorflow
Log in or sign up for Devpost to join the conversation.