Giving machines human-like vision. Edge AI cameras and open software that make Physical AI simple. Building the vision for a better future
Denver, CO
Joined December 2012
- Tweets2.8K
- Following215
- Followers3.6K
- Likes10.3K
Luxonis | Robotic Vision retweeted
Build an OAK app around a VLM in about 15 minutes.
One prompt to a coding agent with the Luxonis Agent Toolkit gets you live video, a scene description every 15 seconds from LFM2.5-VL-450M on OAK4, and a searchable history of every frame it saw.
Luxonis | Robotic Vision retweeted
You can now run vision-language models on OAK4.
Ask the camera if a pallet is fully wrapped and get an answer, generated entirely on-device.
New llama.cpp and ONNX Runtime support means no conversion step for VLMs and other larger models. Examples are live, more coming soon
Luxonis | Robotic Vision retweeted
Custom model training now in Luxonis Hub, Part 2:
Label your data, train, and convert the model for OAK in one workflow. Here's how each step works! 🧵👇
Test with a prompt, a link to the release, and our MCP installed
Don't want to spend $20k+ to capture accurate point clouds of the items in your bin?
Instead, deploy a few OAK 4s running our large Neural Stereo model and fuse their point clouds
Full release: discuss.luxonis.com/blog/706…
Build an OAK app around a VLM in about 15 minutes.
One prompt to a coding agent with the Luxonis Agent Toolkit gets you live video, a scene description every 15 seconds from LFM2.5-VL-450M on OAK4, and a searchable history of every frame it saw.
Full walkthrough, the exact prompt, and model comparison: discuss.luxonis.com/blog/706…
Depth AI 3.11 Release: one pipeline, several OAK 4 cameras.
Add cameras with addDevice(), sync their frames, and calibration them into one coordinate system with the new MultiDeviceCalibration node (beta). No calibration target.
In depth walkthrough and code: discuss.luxonis.com/blog/706…
You can now run vision-language models on OAK4.
Ask the camera if a pallet is fully wrapped and get an answer, generated entirely on-device.
New llama.cpp and ONNX Runtime support means no conversion step for VLMs and other larger models. Examples are live, more coming soon
Custom model training now in Luxonis Hub, Part 2:
Label your data, train, and convert the model for OAK in one workflow. Here's how each step works! 🧵👇
It now trains OCR too.
Draw the text region, type what it says, and train on your own plates, labels or serial numbers. Then convert for OAK and read text directly on-device.
Every team currently working in Hub has access to this and additional promotional training credits each month to try it out, at no extra cost.
Get Started: hub.luxonis.com/create-team?…
Workflow Guide: docs.luxonis.com/cloud/hubai…
Luxonis | Robotic Vision retweeted
Custom model training is now live in Luxonis Hub!
Label your images, train, evaluate, and deploy to OAK. When the model misses something in the field, snap it from the camera and train the next version.
Teams already in hub can try this at no cost.
Read more: discuss.luxonis.com/blog/705…
Custom model training is now live in Luxonis Hub!
Label your images, train, evaluate, and deploy to OAK. When the model misses something in the field, snap it from the camera and train the next version.
Teams already in hub can try this at no cost.
Read more: discuss.luxonis.com/blog/705…
We just released a new customer story on our blog!
Cavalla's autonomous forklifts run with the support of OAK-D Pro W PoE as their main mapping sensor.
IP65 for warehouse dust. Global shutter so vibration doesn't wreck depth. One locking PoE cable for power and data, and stereo depth on-device.
The future of warehousing is here - and it is powered by OAK.
Read full article: discuss.luxonis.com/blog/704…
