Lookout

Inspiration

I noticed that there are thousands of cameras indexed by services such as Shodan. I thought perhaps facial recognition could be used to identify and potentially prevent crimes from happening.

What it does

The program identifies and criminals who have been reported on family watchdog.

How I built it

I scraped information on criminals from Family Watchdog to create database of criminals. Used dlib face recognition model and OpenCV to identify the criminal from the database from Watchdog.

Challenges I ran into

The website I used for information on criminals (familywatchdog.us) near a given location only had a paid API that required contacting them for access, so I had to scrape the data using BeautifulSoup and python requests.
Everyone from Applied Physics Laboratory was very helpful, in helping me overcome this obstacle.

Accomplishments that I'm proud of

Gaining a better understanding of internet protocols and website scrapping. Managed to make real time face recognition much quicker by using haar cascades to first identify the faces, then use dlib's model to recognize faces. Additionally uses no API for recognition unlike many other past projects allowing for greater control and scaling when I have more computers to process data.

What's next for Lookout

Making a GUI and adding functionality to simply connect to other wifi enabled cameras and CCTV cameras already currently in place.

Built With

• python