Inspiration
The world is becoming evermore connected, and learning a language is hard. There are words in some languages that don't exist in others. Other times, we simply do not know what the thing is called. Lingoplex provides a simple way to learn new words in a foreign language by simply taking a picture of the object.
What it does
Lingoplex uses a deep convolutional neural network to identify over thousands of images, trained from data from the 2012 the ImageNet Large Visual Recognition Challenge
How we built it
We utilized Tenorflow API from Google, in which we utilized its functions for image recognition. We developed Python scripts and run the API. We utilized HTML for our application interface, which is connected to Python scripts and will do the image processing and translating for us.
Challenges we ran into
Since we only have less than one day to complete this project, timing issue is the crucial challenge we ran into. We tried to learn new concept/knowledge and improve interface design at the same time, yet we ran into the problem that due to time limit, we cannot achieve both objectives perfectly.
Accomplishments that we're proud of
We are able to finish the project within the time limit, and it can run it without errors.
What we learned
We are able to get a hang of some basis of machine learning, and are able to use some of them onto our project. We also learned some interface design techniques on HTML.
What's next for Lingoplex
We will be continuing on improving the back-end database for words. This is because the database we are currently using is small and is manually generated/typed by our members. We are also going to continuously improving the graphical interface of Lingoplex, in order to make it more attractive and colorful.
Log in or sign up for Devpost to join the conversation.