Viewing all available memories uploaded by users on the website
Clicking on one memory shows only other relevant memories
We human beings all have a limited life span, and time is our only enemy that washes valuable memories away. We find self validation by sharing our sorrows and happiness, by connecting with people with similar experiences. Spreading a cheerful moment of your life creates positive energy throughout our community. When faced with a challenging hurdle in your life, it is vital to connect with others that have been through this and have empathy from others.
What it does
By submitting pieces of memories, the website would automatically analyze the videos for semantic content it presents. Then, it looks for other memories that present a similar story. It presents these connections to the user and hence sharing memories.
How we built it
The backend was programmed in Python. By calling the Google Video Intelligence APIs as well as word2vec, the backend is able to analyze for important topics and also measure the similarity between memories.
Challenges we ran into
It was really hard to read through and fully understand the Google video intelligence API within a limited amount of time. We used cytoscape, a data visualization library for bioinformatics, for front end development to show the linkages between different videos. The learning curve for this library is very steep and some important details are not well documented.
Accomplishments that we're proud of
We have integrated the various data sources, including the machine learning APIs as well as the semantic matching algorithm, to create something seamless and intuitive to the user.
What we learned
To increase our productivity, we have isolated the front/backend such that multiple team members can simultaneously work without interfering with each other. Through this process, we have executed the agile software development process with fast iteration and frequent communication.
What's next for Memoritis
In the future, the process of recording and submitting memories can be automated using a wearable device. With proper user consent, the device is able to capture upload the important moments to be shared with other peers using the service.