search

CVHub520 / X-AnyLabeling

Effortless data labeling with AI support from Segment Anything and other awesome models.
GNU General Public License v3.0
4.26k stars 484 forks source link

SAM2 not producing masks in video mode #666

Closed gboeer closed 1 month ago

gboeer commented 1 month ago

Search before asking

X-AnyLabeling Component

No response

Bug

Hi, I can run the labeling app, load a SAM2 video model and also annotate single images just fine and the produced segmentation masks look fine. However, when I try to perform the video processing, no masks are produced in the subsequent frames.

Each image is processed, according to the progress bar, but afterward no new masks are created.

Information

Application Information: {'App name': 'X-AnyLabeling', 'App version': '2.4.4', 'Device': 'CPU'}

System Information: {'CPU': 'x86_64', 'CUDA': None, 'GPU': '0, NVIDIA GeForce RTX 2080 Ti, 11264, 1, NVIDIA GeForce RTX 2080 Ti, ' '11264, 2, NVIDIA GeForce RTX 2080 Ti, 11264, 3, NVIDIA GeForce RTX ' '2080 Ti, 11264, 4, NVIDIA GeForce RTX 2080 Ti, 11264, 5, NVIDIA ' 'GeForce RTX 2080 Ti, 11264, 6, NVIDIA GeForce RTX 2080 Ti, 11264, 7, ' 'NVIDIA GeForce RTX 2080 Ti, 11264, 8, NVIDIA GeForce RTX 2080 Ti, ' '11264, 9, NVIDIA GeForce RTX 2080 Ti, 11264', 'Operating System': 'Linux-6.8.0-44-generic-x86_64-with-glibc2.31', 'Python Version': '3.10.15'}

Package Information: {'ONNX Runtime GPU Version': '1.18.0', 'ONNX Runtime Version': '1.16.0', 'ONNX Version': '1.13.1', 'OpenCV Contrib Python Headless Version': '4.10.0.84', 'PyQt5 Version': '5.15.7'}

Link to a Reproducible Demonstration Video

none

Execution Mode

Source Code

Additional

Are you willing to submit a PR?

gboeer commented 1 month ago

I think this is related to #613 I will try to fix the _C import error first