Spaces:

hysts
/

ViTPose-transformers

Running on Zero

hysts HF staff commited on 2 days ago

Commit

b31249e

1 Parent(s): de856b7

Fix typo

Files changed (1) hide show

app.py CHANGED Viewed

@@ -55,7 +55,7 @@ def process_image(image: PIL.Image.Image) -> tuple[PIL.Image.Image, list[dict]]:
     inputs = pose_image_processor(image, boxes=[person_boxes], return_tensors="pt").to(device)
-    # for vitpose-plus-base checkpoint we should additionaly provide dataset_index
     # to specify which MOE experts to use for inference
     if pose_model.config.backbone_config.num_experts > 1:
         dataset_index = torch.tensor([0] * len(inputs["pixel_values"]))

     inputs = pose_image_processor(image, boxes=[person_boxes], return_tensors="pt").to(device)
+    # for vitpose-plus-base checkpoint we should additionally provide dataset_index
     # to specify which MOE experts to use for inference
     if pose_model.config.backbone_config.num_experts > 1:
         dataset_index = torch.tensor([0] * len(inputs["pixel_values"]))