Inspiration
Human trafficking is a huge problem for humanity right now. Survivors of human trafficking can provide insights into the operation that victimized them, but only have partial information, which is often insufficient for law enforcement to take any action. By using the power of generative AI and RAG, law enforcement can gain better insights into the logistical infrastructure of these criminal enterprises and crack down on them. The possibility of making an impact on a problem that is affecting millions of lives each year is what inspired me.
What it does
Hound takes survivor testimony to not only generate visual clues from their description, but also bridge any information gaps. For example the survivor stayed in a room with a wallpaper with red roses and was located near a white lighthouse, overlooking a cliff. They can describe what they saw in detail and the AI will generate realistic pictures of the location. The law enforcement can then use the photos to better investigate. Sometimes the victims don't even know where they are. They have pieces of information. Hound can take fragments of information and by searching publicly available information like reviews on Google or social media, to connect the dots and zero in on the exact location where the victim was placed. Since this place is a part of the trafficking infrastructure, the law enforcement can investigate further.
How we built it
I created a basic web app using react js and backed it up with Llama Index based RAG, running on Groq, wrapped by a basic Python FastAPI web server.
Challenges we ran into
Integrating so many technologies and testing them in such a short amount of time was the most challenging part. If something didn't work, debugging it fast was a challenge.
Accomplishments that we're proud of
Albeit unpolished and minimalistic, completing the project in time is something that I am really proud of.
What we learned
I learned a lot about RAG and how to use Llama Index. I had previously thought deploying a RAG application would be trivial but I found that it is pretty complex. Overall it was an amazing learning experience.
Log in or sign up for Devpost to join the conversation.