SkalskiP ✓
@skalskip92 · Kraków, Polska · joined 20 Feb 2014
Open-source Lead @roboflow. VLMs. GPU poor. Dog person. Coffee addict. Dyslexic. | GH: https://t.co/dEmzMDGXVf | HF: https://t.co/4Lx1Yw3CLF
49 787Followers
1 409Following
12 781Posts total
4.1MViews on collected posts
Последние посты
Ищем новые посты…
prompt GPT-6 Astra with boxes to generate new detections
- green - positive examples
- red - negative examples
- blue - new detections https://t.co/zfJY4xbuYs
19.3K views · 218 likes
· 22 reposts · 9 replies
Open on X →
0:04
basketball AI (95% local AI + 5% GPT-6 Astra)
- detect ball and players
- track players
- re-identify players across plays
- OCR player numbers
- recognize player in possession
- detect court keypoints
- map player positions and trajectories
Astra is powerful, but expensive htt
145.4K views · 1.4K likes
· 138 reposts · 55 replies
Open on X →
0:18
prompt GPT-6 Astra with boxes to generate new detections
really useful when you know the object, but can’t describe it well in words (like shipping container wall dent)
best part is you can do this between different images
- green - positive examples
- red - negative examples
32.5K views · 120 likes
· 7 reposts · 12 replies
Open on X →
Astra really feels like “ChatGPT moment” for computer vision
23.1K views · 448 likes
· 11 reposts · 18 replies
Open on X →
@skalskip92 Might be the moment where we crossed the threshold.
- Astra (batch): ~$0.03–0.06 / image
- Human (India/PH, $0.02–0.08 per box × 12–18 objects): $0.25–$1.00 / image
1,000 frames:
- Astra: $30–60
- Human: $250–800
Astra also does the team ID + home/away context in
3.4K views · 35 likes
· 0 reposts · 2 replies
Open on X →
@skalskip92 for anything with a million internet photos behind it, agreed.
the labeling that actually gets paid for is the long tail: 40 examples of one defect on one production line, classes the customer invented last month, zero web presence.
knowing celtics jerseys says noth
5.6K views · 37 likes
· 1 reposts · 2 replies
Open on X →
data labeling is dead
GPT-6 Astra easily tells Celtics players from Knicks players
whether they play at home or on the road https://t.co/RCth36rbiU
642.1K views · 3K likes
· 186 reposts · 74 replies
Open on X →
0:21
@skalskip92 nice, btw for video understanding gemini flash is still the best bet, right?
2K views · 11 likes
· 0 reposts · 1 replies
Open on X →
GPT-6 Astra is fast and expensive https://t.co/EjHb51ooBs
5K views · 43 likes
· 2 reposts · 2 replies
Open on X →
empty / full gas cylinder detection, GPT-6 Astra (high)
how to tell which one is open and which one is closed?
the only clue is the pin on the valve https://t.co/jtzS96BLYk
5.8K views · 36 likes
· 1 reposts · 1 replies
Open on X →
cargo container damage detection https://t.co/MpabGLb5na
4.5K views · 29 likes
· 1 reposts · 1 replies
Open on X →
tire size text detection
GPT-6 Astra sees text everywhere https://t.co/OiuNFWqt5W
4.3K views · 34 likes
· 1 reposts · 1 replies
Open on X →
ceramic insulator damage detection https://t.co/NGi5mBFR7Z
4.8K views · 32 likes
· 1 reposts · 1 replies
Open on X →
poker chips detection https://t.co/Dt0VSNGH6T
5.2K views · 34 likes
· 1 reposts · 1 replies
Open on X →
areal and satellite images https://t.co/uLu9sWCa2q
5.7K views · 35 likes
· 2 reposts · 3 replies
Open on X →
jelly bean detection
notice I did not use color. I used taste. https://t.co/tuhP1zHrG8
6.3K views · 40 likes
· 2 reposts · 1 replies
Open on X →
cargo container serial number https://t.co/1SD4TASZGA
7K views · 38 likes
· 3 reposts · 1 replies
Open on X →
document layout detection https://t.co/fd6cyZVBzM
7.9K views · 44 likes
· 2 reposts · 1 replies
Open on X →
technical drawings https://t.co/Txky7ioF7v
10.2K views · 54 likes
· 3 reposts · 2 replies
Open on X →
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 https://t.
880.9K views · 1.4K likes
· 88 reposts · 48 replies
Open on X →
detection mAP@50
- low effort: #1, 82.1
- high effort: #1, 83.6
link: https://t.co/puUowt3HbY https://t.co/f7dYaFHwCm
18.9K views · 88 likes
· 5 reposts · 1 replies
Open on X →
@skalskip92 is it open source, your hands, or the prompt or testing environment? i want a similar thing and i tried for a week but the architecture and prompt for object detection, bounding box, drawings, and coordinates, etc
74 views · 1 likes
· 0 reposts · 0 replies
Open on X →
@skalskip92 I am beginning to understand why Qwen didn’t release the vision encoder on the new qwen3.8 max! It’s so strong!
741 views · 5 likes
· 0 reposts · 1 replies
Open on X →
Gemini 3.8 Flash’s speed is insane
link: https://t.co/tC22dvHj1Y https://t.co/VP1YePh0dS
1.9K views · 18 likes
· 0 reposts · 2 replies
Open on X →
spatial reasoning, Gemini 3.8 Flash (high)
compare Gemini 3.8 Flash, Qwen 3.8 Flash, Muse Spark 1.3, Claude Fable 5.1 and GPT 5.6 Terra: https://t.co/MuTyzQKuou https://t.co/KeKnDZiQyB
2.1K views · 16 likes
· 0 reposts · 1 replies
Open on X →
technical drawing reasoning, Gemini 3.8 Flash (high) https://t.co/8DP5T5C93h
1.7K views · 13 likes
· 0 reposts · 2 replies
Open on X →
tables with handwritten data reasoning, Gemini 3.8 Flash (high) https://t.co/U9RxDmpHT4
1.6K views · 13 likes
· 0 reposts · 1 replies
Open on X →
quality assurance, Gemini 3.8 Flash (high) https://t.co/CP093s8iGV
1.9K views · 14 likes
· 0 reposts · 1 replies
Open on X →
На фоне аккаунтов своего размера
42 постов за последние 90 дней рядом с диапазоном 10K–100K подписчиков. показывается широко, но откликается мало кто.
Медианные просмотры5 136этот аккаунт924медиана для 10K–100K
Охват, %10.32%этот аккаунт3.62%медиана для 10K–100K
Вовлечённость, %0.69%этот аккаунт1.52%медиана для 10K–100K
| Показатель | Этот аккаунт | Медиана для 10K–100K | Отношение |
|---|---|---|---|
| Медианные просмотры на пост | 5 136 | 924 | 5.56× |
| Охват (просмотры ÷ подписчики) | 10.32% | 3.62% | 2.85× |
| Вовлечённость | 0.69% | 1.52% | 0.46× |
Другие в этом диапазоне → Сравнить с другим аккаунтом → Как считаются эти ориентиры →
Growth & engagement
How the posts we collected actually performed: views and reaction rate post by post, what the audience did with them, and where the follower count goes.
Views per post
4.8K5 Sep
4.3K
4.5K
5.8K
5K
2K
642.1K6 Sep
5.6K
3.7K
3.4K
23.1K7 Sep
32.5K8 Sep
145.4K10 Sep
19.3K11 Sep
Last 14 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.
Engagement rate per post
0.70%5 Sep
0.83%
0.69%
0.69%
0.95%
0.59%
0.51%6 Sep
0.71%
0.62%
1.09%
2.06%7 Sep
0.43%8 Sep
1.12%10 Sep
1.29%11 Sep
Reactions — likes, reposts, replies and quotes — divided by views. Median for 10K–100K accounts is 1.52%.
What the audience does
Likes59.1%7 263 in total
Reposts3.9%477 in total
Replies2.0%247 in total
Quotes0.6%78 in total
Bookmarks34.4%4 234 in total
Share of every reaction we collected for this account. Replies mean argument, reposts mean endorsement, bookmarks mean the post was worth keeping.
The follower curve appears once this account has two daily snapshots — we take one a day, and this one is on its first.
Похожие аккаунты
Mike North ✓
@North2North
99.6K followers
· 182.9K posts
Charles Patterson ✓
@CharlesPattson
99.6K followers
· 21.3K posts
John MacDonald ✓
@JohnnyBoi_i
99.6K followers
· 7K posts
Janusz Walczuk ✓
@walczukjanusz
99.6K followers
· 2.6K posts
CRYPTOKRALI©️ ✓
@CRYPTOKRALI3
99.6K followers
· 153.8K posts
Prep Redzone 🏈 ✓
@PrepRedzone
99.6K followers
· 53.2K posts
Team Saath Official🤝 ✓
@TeamSaath
99.6K followers
· 23.9K posts
Lincoln
@LincolnMotorCo
99.6K followers
· 8.3K posts