Hey YC! I’m Alex, founder of Dragoneye.

I want to share with y’all the launch of our object detection platform, which can recognize 40x more classes than Amazon Rekognition and 20x more than Google Vision across a wide range of categories.

The Problem:

You’re building a new feature that relies on the ability to recognize things in camera or photos. But all of the off-the-shelf APIs today don’t get the detail you need. You don’t just need to know it’s a dress, but that it’s a checkered short-shift dress!

You look into how to train your own computer vision model, but you soon realize what a daunting task it is. There are thousands to millions of photos to collect, and it takes weeks of time. What can you do?

Our Solution:

We want to have the best visual recognition API out there, and enable devs to build powerful vision-driven features in a pinch.

What we’re focused on:

• Latency - our fastest models can detect objects in 12ms
• Detail - 13,000 different classes over 25 categories now, and we aim to get to 100,000+ by the end of 2024
• Implementation ease - under 10 lines in Python & JS and a simple REST API

Seeing is believing - pun intended - so here is our side-by-side comparison with Google’s Vision API and Amazon’s Rekognition

Our Ask

1. Try out our instant demo at https://dashboard.dragoneye.ai. We know we have some bugs to squash, so please let us know what works great and what doesn’t.
2. Book a meeting with us if you have computer vision/object detection needs - coffee is on us!