SAM3 Tracker WebGPU
🎯
20
Segment and extract parts from images by clicking
AI & ML interests
None defined yet.
Recent Activity
View all activity
File upload not working under linux
7
#1 opened 17 days ago
by
nils-deybach
Xenova
updated
a
Space
7 days ago
add-text-preprocessing
3
#2 opened 9 days ago
by
miraclemind
Xenova
updated
a
Space
12 days ago
Xenova
published
a
Space
12 days ago
Xenova
updated
a
Space
12 days ago
Xenova
updated
a
Space
13 days ago
Update demo (Transformers.js v3.8.0)
#4 opened 13 days ago
by
Xenova
Xenova
published
a
Space
17 days ago
Xenova
published
a
Space
19 days ago
Xenova
updated
a
Space
19 days ago
Xenova
updated
a
Space
about 2 months ago
Xenova
published
a
Space
about 2 months ago
Xenova
published
a
Space
2 months ago
Xenova
updated
a
Space
2 months ago
Xenova
updated
2
Spaces
3 months ago
Update files
#2 opened 3 months ago
by
Xenova
Xenova
published
a
Space
3 months ago
Post
11659
Okay this is insane... WebGPU-accelerated semantic video tracking, powered by DINOv3 and Transformers.js! 🤯
Demo (+ source code): webml-community/DINOv3-video-tracking
This will revolutionize AI-powered video editors... which can now run 100% locally in your browser, no server inference required (costs $0)! 😍
How does it work? 🤔
1️⃣ Generate and cache image features for each frame
2️⃣ Create a list of embeddings for selected patch(es)
3️⃣ Compute cosine similarity between each patch and the selected patch(es)
4️⃣ Highlight those whose score is above some threshold
... et voilà! 🥳
You can also make selections across frames to improve temporal consistency! This is super useful if the object changes its appearance slightly throughout the video.
Excited to see what the community builds with it!
Demo (+ source code): webml-community/DINOv3-video-tracking
This will revolutionize AI-powered video editors... which can now run 100% locally in your browser, no server inference required (costs $0)! 😍
How does it work? 🤔
1️⃣ Generate and cache image features for each frame
2️⃣ Create a list of embeddings for selected patch(es)
3️⃣ Compute cosine similarity between each patch and the selected patch(es)
4️⃣ Highlight those whose score is above some threshold
... et voilà! 🥳
You can also make selections across frames to improve temporal consistency! This is super useful if the object changes its appearance slightly throughout the video.
Excited to see what the community builds with it!
- 1 reply