How to use it, if there is a video, how to infer, like aot-benchmark demo？ about dmaot-vots2023

[yamy-cheng]

Yes, you can modify the EngineConfig parameter in tools/demo.py to utilize the dmaot engine for video inference.

[zhanghongyong123456]

Yes, you can modify the EngineConfig parameter in tools/demo.py to utilize the dmaot engine for video inference.

thanks when，i run vot initialize vots2023 --workspace Outputs have this error ：Traceback (most recent call last):
File "H:\Anaconda3\envs\dmaot\lib\runpy.py", line 194, in run_module_as_main
return run_code(code, main_globals, None,
File "H:\Anaconda3\envs\dmaot\lib\runpy.py", line 87, in run_code
exec(code, run_globals)
File "H:\Anaconda3\envs\dmaot\Scripts\vot.exe_main.py", line 4, in
File "H:\Anaconda3\envs\dmaot\lib\site-packages\vot\utilities\cli.py", line 9, in
from ..tracker import Registry, TrackerException
File "H:\Anaconda3\envs\dmaot\lib\site-packages\vot\tracker_init.py", line 14, in
from vot.dataset import Frame
File "H:\Anaconda3\envs\dmaot\lib\site-packages\vot\dataset_init.py", line 11, in
from trax import Region
File "H:\Anaconda3\envs\dmaot\lib\site-packages\trax_init_.py", line 7, in
from ._ctypes import
File "H:\Anaconda3\envs\dmaot\lib\site-packages\trax_ctypes.py", line 503, in
_libs["libtrax"] = load_library("libtrax")
File "H:\Anaconda3\envs\dmaot\lib\site-packages\trax_ctypes.py", line 191, in call
raise ImportError("Could not load %s." % libname)
ImportError: Could not load libtrax.
The only difference in my environment is the different versions of torch（default： torch==1.10.1+cu102 My： torch==1.13.0+cu117（my cuda：11.7））

[yamy-cheng]

It seems that there is an issue with the vot-toolkit. Would you like to reinstall it?

[zhanghongyong123456]

It seems that there is an issue with the vot-toolkit. Would you like to reinstall it?

OK，thanks，Which is easier to succeed with pip installation or git clone source code installation

[zhanghongyong123456]

Yes, you can modify the EngineConfig parameter in tools/demo.py to utilize the dmaot engine for video inference.
I'm not sure which one is right，

[yamy-cheng]

Yes, you can modify the EngineConfig parameter in tools/demo.py to utilize the dmaot engine for video inference.
I'm not sure which one is right，

You can pull the latest version of the code and use the following command to perform video inference using r50_dm_deaot. python tools/demo.py --model r50_dm_deaotl --ckpt_path path for r50-DeAOT --max_long 10.

[zhanghongyong123456]

Yes, you can modify the EngineConfig parameter in tools/demo.py to utilize the dmaot engine for video inference.
I'm not sure which one is right，

You can pull the latest version of the code and use the following command to perform video inference using r50_dm_deaot. python tools/demo.py --model r50_dm_deaotl --ckpt_path path for r50-DeAOT --max_long 10.

thanks,It's OK