Closed AgentEXPL closed 2 years ago
@ykarmesh
@AgentEXPL - Yes. The index has changed. You can check this in the episode datasets released for the corresponding challenges. In the val/val.json.gz
dataset from challenge 2021 (with MP3D dataset), this was the mapping from objects to task ids:
"category_to_task_category_id": {"chair": 0, "table": 1, "picture": 2, "cabinet": 3, "cushion": 4, "sofa": 5, "bed": 6, "chest_of_drawers": 7, "plant": 8, "sink": 9, "toilet": 10, "stool": 11, "towel": 12, "tv_monitor": 13, "shower": 14, "bathtub": 15, "counter": 16, "fireplace": 17, "gym_equipment": 18, "seating": 19, "clothes": 20}
In the current 2022 challenge, this is the mapping from val/val.json
:
"category_to_task_category_id": {"chair": 0, "bed": 1, "plant": 2, "toilet": 3, "tv_monitor": 4, "sofa": 5}
I hope this answers your question.
A new dataset called HM3D-Semantics v0.1 is used in ObjectNav Challenge 2022. Compared with the old dataset used in Challenge 2021, whether the number index indicating the category of object is changed in the new dataset? Is there any methods/codes to check this.