Build & use computer vision models fast ✨ Get started: roboflow.com/ Conference Oct 22 in SF: summit.roboflow.com/

Roboflow retweeted
Opus 5.5 is the best "vision" model Anthropic ever made better than Fable 5, better than GPT-6 Sol, worse than GPT-6 Astra vs Fable 5.1 (high): - estimated cost: ↓ 60% - detection: ↑ 11.84 pp - reasoning: ↑ 12.80 pp vs GPT-6 Sol (high): - extraction: ↑ 11.00 pp - reasoning: ↑ 7.95 pp ↓ 16 more examples
fuck me. every lab is working on vision. I have preliminary Opus 5.5 results. by far the best vision model from Anthropic. more soon.
22
26
382
34,965
Roboflow retweeted
going live tomorrow to test GPT-6 Sol and Luna, Opus 5.5, MiMo-V2.6 Pro & Flash and Grok 4.7 Sep 24th, 17:00 CEST / 11:00 EDT / 08:00 PDT link: piped.video/live/E57bsVJtHT8
1
1
22
1,534
Roboflow retweeted
Ok this is pretty insane. Astra side-loaded a custom app for my DJI RC2 controller that takes a live video stream from my Mini 5 Pro drone, streams it to a @roboflow Workflow running RF-DETR via our Serverless Video API, & overlays the predictions back on the controller’s screen.
41
51
768
59,574
Roboflow retweeted
Opus 5.5, GPT-6 Sol, and GPT-6 Luna numbers are on the Playground leaderboard crazy how fast Google lost the computer vision lead object detection used to be their stronghold, now 3 labs and 4 models sit ahead of them link: playground.roboflow.com/eval…
10
15
169
10,900
can you train an object tracker on the fly brain? yes! meet robofly 🪰 had fun hacking around to see how the fly brain lights up related to different visual inputs the difference between single object tracking and multiple things moving is really cool to see link to try it out below open to ideas and suggestions on where we should take this
4
20
1,608
we've started to evaluate Astra for segmentation tasks and it's really good but as we all know it's really expensive, ~$0.50 for a 1920 × 1280 image for example we think the best way to do zero-shot segmentation is to you use Astra on low/medium for object detection and then pass that to SAM3 you can get pixel perfect masks for 80% cheaper than having Astra draw polygons Astra proposes boxes, SAM3 draws the masks, and you get Astra's understanding with SAM3's edges at lower cost per image, because boxes are less complex than polygon lines, less Astra's output tokens
SAM3 vs Astra for text-prompt segmentation SAM3 gives much more precise masks. it just often fails to understand complicated text prompts. Astra is great at language. it rarely misses what you want to detect, usually only when the task needs internal domain knowledge. masks come back as JSON, point by point. slower, and less accurate than SAM3 we just rolled out Astra segmentation in playground link: playground.roboflow.com/shar… ↓ more Astra examples below
3
12
151
10,718
Speaker lineup and agenda announced for the Visual Intelligence Summit Going to be an intense day with 30+ speakers covering topics across industries, from robotics to frontier models, edge vision models to powerful agents running in the cloud Hands on workshops on - Model training and architectures - Frontier and multimodal models - Robotics and edge deployments - Data collection, curation, and labeling For anyone building in the vision space, this is the day to build your network and learn the latest advancements as you work at the frontier of what's possible with vision
We are hosting the Visual Intelligence Summit on October 22 in SF. The physical world is AI's biggest opportunity and the best way to advance humanity in our lifetimes. Few people are building for AI beyond the screen. The Summit is for the people who see it differently and want to be at the frontier of bringing AI into the physical world. If you’re building for that future, join us.
1
10
1,324
Astra built a complex breathing detection application This video actually steps through the front-end to show you how Astra built it, not just a demo. He gave Astra a stack of videos of himself breathing and let it figure out the rest. Astra ran five experiments on its own to come to the best conclusion. We'll be sharing more behind the scenes on how Astra demos are being built
New tutorial on how to use Astra for training smaller models Astra is powerful but too expensive for many use cases most Astra demos don't make sense as production applications so we made a tutorial showing how you can use Astra's vision capabilities to train smaller more efficient models like RF-DETR Preview here, full tutorial in next post
2
39
2,766
New tutorial on how to use Astra for training smaller models Astra is powerful but too expensive for many use cases most Astra demos don't make sense as production applications so we made a tutorial showing how you can use Astra's vision capabilities to train smaller more efficient models like RF-DETR Preview here, full tutorial in next post
we've been testing astra across domains and are finding some surprises and specific areas it is reliable exciting zero-shot knowledge wins happening here, mainly understanding which player scored! a lot goes into that: - knowing in/out of court and service box - who hit last - understanding a point ended - game knowledge of missed first-serve doesn't result in a point other cool things here you get out of the box - understanding the stroke - tracking player to assign points would be easy to add - stroke counts - heat map - point durations
4
15
186
14,129
end to end data labeling tutorial with Astra
2
5
676
we've been testing astra across domains and are finding some surprises and specific areas it is reliable exciting zero-shot knowledge wins happening here, mainly understanding which player scored! a lot goes into that: - knowing in/out of court and service box - who hit last - understanding a point ended - game knowledge of missed first-serve doesn't result in a point other cool things here you get out of the box - understanding the stroke - tracking player to assign points would be easy to add - stroke counts - heat map - point durations
19
25
461
58,004
great model to create a quick prototype and prove an idea can be expensive depending on your use case
1
3
1,718
the Driveline team is pushing the frontier of what's possible with AI + sports if you're interested in how AI is shaping the future of sports, you have to come check out their session
Driveline Baseball x Roboflow Come see @SamEhrlich and I speak in San Francisco at the @roboflow Visual Intelligence Summit on October 22nd. @DrivelineBB is pushing the boundaries of applied AI/CV in sports and we're honored to be selected for VIS! summit.roboflow.com/
3
5
49
7,981
you can label data in roboflow with astra millions of vision products are now unblocked by labeling time/cost going almost to $0 crazy next 12 months ahead
data labeling is dead GPT-6 Astra easily tells Celtics players from Knicks players whether they play at home or on the road
7
39
479
39,576
GPT-6 Astra is now the top model for vision tasks sure 3-10x more expensive but 50%+ faster extremely capable model, OpenAI really pushed the frontier in all directions with this one
Gemini dethroned GPT-6 Astra is the best "vision" model I've seen - notices tiny details - draws conclusions - produces precise and tight boxes - sees and reads text in different contexts - knows a lot about a lot - fast - expensive ↓ GPT-6 Astra detection deep dive
6
9
122
8,411