SkalskiP
@skalskip92
Followers
42K
Following
21K
Media
2K
Statuses
10K
Open-source Lead @roboflow. VLMs. GPU poor. Dog person. Coffee addict. Dyslexic. | GH: https://t.co/dEmzMDGq5H | HF: https://t.co/4Lx1Yw34W7
Kraków, Polska
Joined February 2014
football AI code is finally open-source - player detection and tracking - team clustering - camera calibration I still need to work on README; don't judge me on that code: https://t.co/VK0RQFX22z
161
1K
9K
yo, nba people reached out to me about this tutorial.
basketball AI YouTube tutorial is finally live over 1000 hours of work compressed into 37 minutes link below; like and comments please
31
26
1K
I am dethroned as the roboflow chef demo officer; its crazy what you can build with just language, no data labeling
i trained a computer vision model to count chinups roboflow rapid + SAM3 detected the chinup bar - hands - head through natural language prompts then I wrote a python script to count completed reps (head above bar) and annotate the video idea for a gamified fitness app?
7
22
506
next up, tenis / paddle / pickleball?
basketball AI YouTube tutorial is finally live over 1000 hours of work compressed into 37 minutes link below; like and comments please
38
13
293
any good agents course you could recommend? preferably vision agents.
13
0
72
data labeling is dead. long live distillation. from data to object detection endpoint in 90 seconds. link: https://t.co/ss4yC889KT
26
126
1K
I'm working on plugging Qwen3-VL and SAM3 into Basketball AI; looks promissing - Qwen3-VL - identifies what teams are playing - SAM3 - tracks players using text prompts
basketball AI YouTube tutorial is finally live over 1000 hours of work compressed into 37 minutes link below; like and comments please
27
64
782
- player detection with RF-DETR - player tracking with SAM2 - team clustering with SigLIP and K-means - number recognition with SmolVLM2
7
10
179
basketball AI YouTube tutorial is finally live over 1000 hours of work compressed into 37 minutes link below; like and comments please
81
229
3K
still working on visualizations; team clustering
8
15
263
upcoming basketball AI tutorial will be packed with strong visuals; can't wait to ship it
this might be the coolest blogpost I ever written I dove deep into: - player detection with RF-DETR - player tracking with SAM2 - team clustering with SigLIP and K-means - number recognition with SmolVLM2 and ResNet I hope you'll like it link: https://t.co/PPPGTD8L2v
29
79
1K
does it make sense?
this might be the coolest blogpost I ever written I dove deep into: - player detection with RF-DETR - player tracking with SAM2 - team clustering with SigLIP and K-means - number recognition with SmolVLM2 and ResNet I hope you'll like it link: https://t.co/PPPGTD8L2v
26
59
1K
so cool to see people build on top of my tutorials computer vision + motorsports is sucha a cool idea!
Had a lot of fun building this model An ML-powered tool that can analyze and optimize racing lines, compare strategies across drivers, and deconstruct strategic decisions at race starts Loads to do still but progress! Stay tuned! #F1 #IndyCar #Motorsport
10
44
613
SAM3 release kind of derailed the basketball AI release I needed a short break, but I am back in business now 1000 hours of work packed into 50 minutes of video
6
9
231
upload video or images, use SAM3 to auto-annotate with text, distill into smaller model and deploy locally or over custom endpoint link:
2
1
38
- if SAM3 is so good, do we still need other models? - yes. SAM3 is too large for low compute edge devices and too slow for real-time use. we still need light and fast models for edge workloads. distill SAM3 into smaller model and deploy it in seconds with @roboflow rapid
SAM3 video tracking is so good yesterday: collect data, train custom object detector, use tracker to estimate object motion - days today: track anything with text prompt - seconds
21
73
974
SAM3 is open-source model. You can use the models in commercial. You can modify or fine tune. You keep ownership of your modifications. You do not need to release your source code.
SAM3 video tracking is so good yesterday: collect data, train custom object detector, use tracker to estimate object motion - days today: track anything with text prompt - seconds
19
76
1K
SAM3 live stream starts in 15 minutes I'll be mostly coding, and talking about image and video segmentation, text and visual prompts, dataset auto-labeling, and more https://t.co/oVI038RgKS
SAM3 live stream tomorrow; image and video segmentation, text and visual prompts, live coding Nov 20th 4:30 PM CET 10:30 AM ET https://t.co/9OnvF36FwS
2
30
308
manual data labeling is over
SAM3 video tracking is so good yesterday: collect data, train custom object detector, use tracker to estimate object motion - days today: track anything with text prompt - seconds
46
254
3K