how to detect image objects #307
Description
i only need to detect image objects, i use the checkpoint: down from https://huggingface.co/OpenGVLab/InternImage/blob/main/cascade_internimage_xl_fpn_3x_coco.pth, and set config= detection/configs/coco/cascade_internimage_xl_fpn_3x_coco.py. then i try to send a path of a image,
I run the command with: python detection/image_demo.py images/video01_0003.jpg detection/configs/coco/cascade_internimage_xl_fpn_3x_coco.py weights/cascade_internimage_xl_fpn_3x_coco.pth.
i get the message:
D:\software\anaconda\envs\whisper\lib\site-packages\mmdet\apis\inference.py:51: UserWarning: Class names are not saved in the checkpoint's meta data, use COCO classes by default.
warnings.warn('Class names are not saved in the checkpoint's '
D:\software\anaconda\envs\whisper\lib\site-packages\mmdet\datasets\utils.py:66: UserWarning: "ImageToTensor" pipeline is replaced by "DefaultFormatBundle" for batch inference. It is recommended to manually replace it in the test data pipeline in your config file.