Switch to bus image with matching classes

README.md

@@ -50,7 +50,7 @@ transform = transforms.Compose([
     transforms.ToTensor(),
 ])
 
-url = "https://raw.githubusercontent.com/google-deepmind/tips/main/scenic/images/example_image_2.jpg"
+url = "https://huggingface.co/spaces/google/tipsv2-gpu-explorer/resolve/main/examples/zeroseg/pascal_context_00049_image.png"
 image = Image.open(requests.get(url, stream=True).raw)
 pixel_values = transform(image).unsqueeze(0)
 out = model.encode_image(pixel_values)
@@ -62,7 +62,7 @@ print(out.patch_tokens.shape)  # (1, 1024, 1536) — per-patch spatial features
 ### Encode text
 
 ```python
-text_emb = model.encode_text(["a photo of a cat", "a photo of a dog"])
+text_emb = model.encode_text(["a photo of a bus", "a photo of a dog"])
 print(text_emb.shape)  # (2, 1536) — one embedding per query
 ```
 
@@ -71,17 +71,17 @@ print(text_emb.shape)  # (2, 1536) — one embedding per query
 ```python
 import torch.nn.functional as F
 
-classes = ["cat", "dog", "car"]
+classes = ["bus", "car", "dog", "cat"]
 cls = F.normalize(out.cls_token[:, 0, :], dim=-1)
 text_emb = F.normalize(model.encode_text(classes), dim=-1)
 similarity = cls @ text_emb.T
-print(classes[similarity.argmax()])  # cat — predicted class
+print(classes[similarity.argmax()])  # bus — predicted class
 ```
 
 ### Zero-shot segmentation
 
 ```python
-classes = ["cat", "carpet", "floor", "furniture"]
+classes = ["bus", "snow", "mountain", "house", "road"]
 patch_feats = F.normalize(out.patch_tokens, dim=-1)
 text_emb = F.normalize(model.encode_text(classes), dim=-1)
 seg_map = (patch_feats @ text_emb.T).reshape(32, 32, len(classes)).argmax(dim=-1)