Created attachment 984 [details] ffmpeg失败模型 Description of problem: 镜像默认未安装ffmpeg包，导致通义实验室8个CV相关的官方模型无法直接运行，ModelScope社区Ubuntu镜像默认安装了ffmpeg无该问题。 影响模型列表见附件，包括cv_raft_video-frame-interpolation等模型 Version-Release number of selected component (if applicable): 镜像地址：registry.openanolis.cn/openanolis/modelscope:1.10.0-an8 Steps to Reproduce: 1. 启动容器docker run -it registry.openanolis.cn/openanolis/modelscope:1.10.0-an8 2. 根据ModelScope官网，找到模型对应的代码范例（code example），以cv_raft_video-frame-interpolation为例： from modelscope.pipelines import pipeline from modelscope.utils.constant import Tasks from modelscope.outputs import OutputKeys video = 'https://modelscope.oss-cn-beijing.aliyuncs.com/test/videos/video_frame_interpolation_test.mp4' video_frame_interpolation_pipeline = pipeline(Tasks.video_frame_interpolation, 'damo/cv_raft_video-frame-interpolation') result = video_frame_interpolation_pipeline(video)[OutputKeys.OUTPUT_VIDEO] print('pipeline: the output video path is {}'.format(result)) 3. 执行代码范例 Actual results: 以cv_raft_video-frame-interpolation为例，执行报错： 2024-01-23 13:09:13,206 - modelscope - INFO - load model done. 2024-01-23 13:09:13,209 - modelscope - WARNING - No preprocessor field found in cfg. 2024-01-23 13:09:13,209 - modelscope - WARNING - No val key and type key found in preprocessor domain of configuration.json file. 2024-01-23 13:09:13,209 - modelscope - WARNING - Cannot find available config to build preprocessor at mode inference, current config: {'model_dir': '/root/.cache/modelscope/hub/damo/cv_raft_video-frame-interpolation'}. trying to build by task and model information. 2024-01-23 13:09:13,209 - modelscope - WARNING - No preprocessor key ('video-frame-interpolation', 'video-frame-interpolation') found in PREPROCESSOR_MAP, skip building preprocessor. 2024-01-23 13:09:13,252 - modelscope - INFO - load video frame-interpolation done 2024-01-23 13:09:13,264 - modelscope - WARNING - task video-frame-interpolation input definition is missing sh: ffmpeg: command not found Traceback (most recent call last): File "/cache/integration_tests/workspace/test_cv_raft_video-frame-interpolation.py", line 8, in result = video_frame_interpolation_pipeline(video)[OutputKeys.OUTPUT_VIDEO] File "/opt/conda/lib/python3.8/site-packages/modelscope/pipelines/base.py", line 219, in __call__ output = self._process_single(input, *args, **kwargs) File "/opt/conda/lib/python3.8/site-packages/modelscope/pipelines/base.py", line 258, in _process_single out = self.postprocess(out, **postprocess_params) File "/opt/conda/lib/python3.8/site-packages/modelscope/pipelines/cv/video_frame_interpolation_pipeline.py", line 605, in postprocess assert os.system( AssertionError: ffmpeg is not installed correctly! 备注：每个模型报错calltrace有差异，但最后均为ffmpeg的问题 Expected results: 执行不报错 Additional info: ModelScope社区Ubuntu镜像（registry.cn-hangzhou.aliyuncs.com/modelscope-repo/modelscope:ubuntu22.04-cuda11.8.0-py310-torch2.1.0-tf2.14.0-1.10.0 ）默认安装了ffmpeg root@6610060b5bc0:/# conda list |grep ffmpeg ffmpeg 1.4 pypi_0 pypi ffmpeg-python 0.2.0 pypi_0 pypi imageio-ffmpeg 0.4.9 pyhd8ed1ab_0 conda-forge