Discovery Bot

Discovery Bot can identify and create alternative text for any images uploaded on your Discord server!
An example of Discovery Bot's character recognition to create transcription for any image, even bad memes
Discovery Bot can also identify landmarks such as Brooklyn College!
Join the demonstration server and try it out for yourself - Discovery Bot is online 24/7.

**Try Discovery Bot for yourself! [Click here to join the demo server!](https://discordapp.com/invite/x5Kf2pS)**

Inspiration

After learning that Hack Brooklyn was going virtual, I started thinking of fun, inclusive ideas where everyone can get involved and try out my project. I decided to build an accessibility application after noticing how, more than ever, we are dependent on online communication tools that lack features to assist those with disabilities. I decided to choose Discord as the initial platform for Discovery Bot as Hack Brooklyn was using it for their "event space". This way, everyone would already have Discord accounts to easily join in and demo the bot.

What it does

Discovery Bot is incredibly simple to use and requires no commands on the user's part. Once the bot is added to a Discord server, it will wait until an image is uploaded by a user. When a user sends an image, Discovery Bot will immediately jump into action and provide alternative text for the image, transcribe the image if any text is detected, and it will also do it’s best to create a proper description for the image.

How I built it

Discovery Bot was built using Javascript. It utilizes Node.js and the Discord API to get the bot running. I used the Google Cloud Vision API for image processing and recognition. The bot is hosted on the Heroku Cloud Platform so it can be online 24/7.

Challenges I ran into

Several! I never worked with either the Discord API or Google Vision API before so it took a lot of digging through documentation to learn the ropes. Heroku also proved challenging because I had trouble keeping my Google API key secured while the project code was made public.

This was also my first "virtual" Hackathon - and one where I chose to go solo in building the project. I had to depend on myself not to get distracted (while remembering to take a break once in a while and limit my caffeine intake).

Accomplishments that I'm proud of

Getting everything to work! There's a lot of room for improvement, but I'm happy with the finished project and I look forward to seeing everyone trying out the bot and enjoying it's functionalities (whether it be a bug or feature)!

What I learned

To stop looking at pictures of food when I'm hungry. Also, learning how simple getting both the Discord API and Google Vision API working in your project once you learn the ropes. I was pleasantly surprised by how quickly I was able to get everything set up and running!

What's next for Discovery Bot

The reaction buttons for the bot were not 100% implemented in time, but I hope to get that running soon so I can train my own model in recognizing images in the future. I would also love to expand the bot to be compatible with more platforms, such as Slack and Twitter.

(If you made it this far, thanks for looking! Type the command !surprise in the demonstration server for a special message)