LVIS Models not working for Instance Segmentation

I am using the following command to create a **LVISv0.5 Instance Segmentation video** using the LVISv0.5 configs:

`python demo.py --config-file ../configs/LVISv0.5-InstanceSegmentation/mask_rcnn_X_101_32x8d_FPN_1x.yaml --video-input vid1.mp4 --output LVISv05-InstanceSegmentation_mask_rcnn_X_101_32x8d_FPN_1x.mp4`

Now this is not performing any type of segmentation in the output video, whereas when I use **COCO Instance Segmentation on the same video, it works**. The COCO commands is as follows:

`python demo.py --config-file ../configs/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x.yaml --video-input vid2.mp4 --output COCO-InstanceSegmentation_mask_rcnn_R_50_FPN_3x.mp4 --opts MODEL.WEIGHTS detectron2://COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x/137849600/model_final_f10217.pkl`

This is the input video link for your reference:
https://drive.google.com/file/d/1iLqNdLBGUryIl4lAot1YpZOU6s76FUgI/view?usp=share_link

Am I missing something?

Start with demo.py and compare the LVISv0.5 and COCO configuration paths shown in the commands, including how model weights are selected. Reproduce the video inference, then verify whether the LVIS run loads the intended weights and produces instance masks; the issue is resolved when LVIS segmentation appears in the output video.

Written by the indexing model from the issue text.