L
Laya Vision
A 256M model that answers typed questions about an image with calibrated odds
Emerged finds new products worth a look, daily — what’s new today
BlindSight
Webcam person detection where the cloud server never sees a single frame
It is a runnable answer to whether you can put camera inference in the cloud without the cloud seeing the camera, with honest numbers on what that costs today.
BlindSight encrypts each webcam frame on your laptop, sends it to a GPU that runs a small CNN on the ciphertext, and decrypts the verdict locally. Only the client holds the key. The numbers are candid: a few hundred milliseconds of GPU work, several seconds per round trip, and 55 of 64 sample frames right, matching the plaintext model. You need a Belfort eML account; new ones get 100 GPU minutes. Apache-2.0, from Belfort Labs in Belgium.
More like this
Goethe
Turn a Linux box with NVIDIA GPUs into a private, OpenAI-compatible endpoint
Jinfer
An AI inference engine that runs inside the JVM, with no Python sidecar
ContentCam
One tap on your phone gives you a vertical and a wide video, both full frame
Chert FaceTime
Open-source CLI and SDK that puts your LiveKit video agent on a FaceTime call
BLA
Hardens your shipped builds so AI reverse-engineering agents waste effort