Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -65,27 +65,91 @@ def apply_mask(image: np.ndarray, mask: np.ndarray, color: Tuple[float, float, f
|
|
65 |
masked_image = image * (1 - alpha * mask) + alpha * mask * color[np.newaxis, np.newaxis, :] * 255
|
66 |
return masked_image.astype(np.uint8)
|
67 |
|
68 |
-
|
69 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
70 |
model, extractor = load_model(model_name)
|
71 |
attention_maps = process_image(image, model, extractor)
|
72 |
-
num_prefix_tokens = getattr(model, 'num_prefix_tokens', 0)
|
73 |
|
|
|
|
|
|
|
74 |
# Convert PIL Image to numpy array
|
75 |
image_np = np.array(image)
|
76 |
|
77 |
# Create visualizations
|
78 |
visualizations = []
|
|
|
79 |
for layer_name, attn_map in attention_maps.items():
|
80 |
print(f"Attention map shape for {layer_name}: {attn_map.shape}")
|
81 |
-
|
82 |
-
|
83 |
-
|
84 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
85 |
# Reshape the attention map to 2D
|
86 |
-
num_patches = int(
|
87 |
attn_map = attn_map.reshape(num_patches, num_patches)
|
88 |
-
|
89 |
# Interpolate to match image size
|
90 |
attn_map = torch.tensor(attn_map).unsqueeze(0).unsqueeze(0)
|
91 |
attn_map = F.interpolate(attn_map, size=(image_np.shape[0], image_np.shape[1]), mode='bilinear', align_corners=False)
|
@@ -116,18 +180,54 @@ def visualize_attention(image: Image.Image, model_name: str) -> List[Image.Image
|
|
116 |
visualizations.append(vis_image)
|
117 |
plt.close(fig)
|
118 |
|
119 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
120 |
|
121 |
# Create Gradio interface
|
122 |
iface = gr.Interface(
|
123 |
fn=visualize_attention,
|
124 |
inputs=[
|
125 |
gr.Image(type="pil", label="Input Image"),
|
126 |
-
gr.Dropdown(choices=get_attention_models(), label="Select Model")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
127 |
],
|
128 |
-
|
129 |
-
title="Attention Map Visualizer for timm Models. NOTE: This is a WIP.",
|
130 |
description="Upload an image and select a timm model to visualize its attention maps."
|
131 |
)
|
132 |
|
133 |
-
iface.launch(
|
|
|
65 |
masked_image = image * (1 - alpha * mask) + alpha * mask * color[np.newaxis, np.newaxis, :] * 255
|
66 |
return masked_image.astype(np.uint8)
|
67 |
|
68 |
+
|
69 |
+
def rollout(attentions, discard_ratio, head_fusion, num_prefix_tokens=1):
|
70 |
+
# based on https://github.com/jacobgil/vit-explain/blob/main/vit_rollout.py
|
71 |
+
result = torch.eye(attentions[0].size(-1))
|
72 |
+
with torch.no_grad():
|
73 |
+
for attention in attentions:
|
74 |
+
if head_fusion.startswith('mean'):
|
75 |
+
# mean_std fusion doesn't appear to make sense with rollout
|
76 |
+
attention_heads_fused = attention.mean(dim=0)
|
77 |
+
elif head_fusion == "max":
|
78 |
+
attention_heads_fused = attention.amax(dim=0)
|
79 |
+
elif head_fusion == "min":
|
80 |
+
attention_heads_fused = attention.amin(dim=0)
|
81 |
+
else:
|
82 |
+
raise ValueError("Attention head fusion type Not supported")
|
83 |
+
|
84 |
+
# Discard the lowest attentions, but don't discard the prefix tokens
|
85 |
+
flat = attention_heads_fused.view(-1)
|
86 |
+
_, indices = flat.topk(int(flat.size(-1 )* discard_ratio), -1, False)
|
87 |
+
print(indices)
|
88 |
+
print(indices.shape)
|
89 |
+
indices = indices[indices >= num_prefix_tokens]
|
90 |
+
flat[indices] = 0
|
91 |
+
|
92 |
+
I = torch.eye(attention_heads_fused.size(-1))
|
93 |
+
a = (attention_heads_fused + 1.0 * I) / 2
|
94 |
+
a = a / a.sum(dim=-1)
|
95 |
+
result = torch.matmul(a, result)
|
96 |
+
|
97 |
+
# Look at the total attention between the prefix tokens (usually class tokens)
|
98 |
+
# and the image patches
|
99 |
+
# FIXME this is token 0 vs non-prefix right now, need to cover other cases (> 1 prefix, no prefix, etc)
|
100 |
+
mask = result[0, num_prefix_tokens:]
|
101 |
+
width = int(mask.size(-1) ** 0.5)
|
102 |
+
mask = mask.reshape(width, width).numpy()
|
103 |
+
mask = mask / np.max(mask)
|
104 |
+
return mask
|
105 |
+
|
106 |
+
|
107 |
+
def visualize_attention(
|
108 |
+
image: Image.Image,
|
109 |
+
model_name: str,
|
110 |
+
head_fusion: str,
|
111 |
+
discard_ratio: float,
|
112 |
+
) -> Tuple[List[Image.Image], Image.Image]:
|
113 |
+
"""Visualize attention maps and rollout for the given image and model."""
|
114 |
model, extractor = load_model(model_name)
|
115 |
attention_maps = process_image(image, model, extractor)
|
|
|
116 |
|
117 |
+
# FIXME handle wider range of models that may not have num_prefix_tokens attr
|
118 |
+
num_prefix_tokens = getattr(model, 'num_prefix_tokens', 1) # Default to 1 class token if not specified
|
119 |
+
|
120 |
# Convert PIL Image to numpy array
|
121 |
image_np = np.array(image)
|
122 |
|
123 |
# Create visualizations
|
124 |
visualizations = []
|
125 |
+
attentions_for_rollout = []
|
126 |
for layer_name, attn_map in attention_maps.items():
|
127 |
print(f"Attention map shape for {layer_name}: {attn_map.shape}")
|
128 |
+
attn_map = attn_map[0] # Remove batch dimension
|
129 |
+
|
130 |
+
attentions_for_rollout.append(attn_map)
|
131 |
+
|
132 |
+
attn_map = attn_map[:, :, num_prefix_tokens:] # Remove prefix tokens for visualization
|
133 |
+
|
134 |
+
if head_fusion == 'mean_std':
|
135 |
+
attn_map = attn_map.mean(0) / attn_map.std(0)
|
136 |
+
elif head_fusion == 'mean':
|
137 |
+
attn_map = attn_map.mean(0)
|
138 |
+
elif head_fusion == 'max':
|
139 |
+
attn_map = attn_map.amax(0)
|
140 |
+
elif head_fusion == 'min':
|
141 |
+
attn_map = attn_map.amin(0)
|
142 |
+
else:
|
143 |
+
raise ValueError(f"Invalid head fusion method: {head_fusion}")
|
144 |
+
|
145 |
+
# Use the first token's attention (usually the class token)
|
146 |
+
# FIXME handle different prefix token scenarios
|
147 |
+
attn_map = attn_map[0]
|
148 |
+
|
149 |
# Reshape the attention map to 2D
|
150 |
+
num_patches = int(attn_map.shape[0] ** 0.5)
|
151 |
attn_map = attn_map.reshape(num_patches, num_patches)
|
152 |
+
|
153 |
# Interpolate to match image size
|
154 |
attn_map = torch.tensor(attn_map).unsqueeze(0).unsqueeze(0)
|
155 |
attn_map = F.interpolate(attn_map, size=(image_np.shape[0], image_np.shape[1]), mode='bilinear', align_corners=False)
|
|
|
180 |
visualizations.append(vis_image)
|
181 |
plt.close(fig)
|
182 |
|
183 |
+
# Calculate rollout
|
184 |
+
rollout_mask = rollout(attentions_for_rollout, discard_ratio, head_fusion, num_prefix_tokens)
|
185 |
+
|
186 |
+
# Create rollout visualization
|
187 |
+
fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(20, 10))
|
188 |
+
|
189 |
+
# Original image
|
190 |
+
ax1.imshow(image_np)
|
191 |
+
ax1.set_title("Original Image")
|
192 |
+
ax1.axis('off')
|
193 |
+
|
194 |
+
# Rollout overlay
|
195 |
+
rollout_mask_pil = Image.fromarray((rollout_mask * 255).astype(np.uint8))
|
196 |
+
rollout_mask_resized = np.array(rollout_mask_pil.resize((image_np.shape[1], image_np.shape[0]), Image.BICUBIC)) / 255.0
|
197 |
+
masked_image = apply_mask(image_np, rollout_mask_resized, color=(1, 0, 0)) # Red mask
|
198 |
+
ax2.imshow(masked_image)
|
199 |
+
ax2.set_title('Attention Rollout')
|
200 |
+
ax2.axis('off')
|
201 |
+
|
202 |
+
plt.tight_layout()
|
203 |
+
|
204 |
+
# Convert plot to image
|
205 |
+
fig.canvas.draw()
|
206 |
+
rollout_image = Image.frombytes('RGB', fig.canvas.get_width_height(), fig.canvas.tostring_rgb())
|
207 |
+
plt.close(fig)
|
208 |
+
|
209 |
+
return visualizations, rollout_image
|
210 |
+
|
211 |
|
212 |
# Create Gradio interface
|
213 |
iface = gr.Interface(
|
214 |
fn=visualize_attention,
|
215 |
inputs=[
|
216 |
gr.Image(type="pil", label="Input Image"),
|
217 |
+
gr.Dropdown(choices=get_attention_models(), label="Select Model"),
|
218 |
+
gr.Dropdown(
|
219 |
+
choices=['mean_std', 'mean', 'max', 'min'],
|
220 |
+
label="Head Fusion Method",
|
221 |
+
value='mean' # Default value
|
222 |
+
),
|
223 |
+
gr.Slider(0, 1, 0.9, label="Discard Ratio", info="Ratio of lowest attentions to discard")
|
224 |
+
],
|
225 |
+
outputs=[
|
226 |
+
gr.Gallery(label="Attention Maps"),
|
227 |
+
gr.Image(label="Attention Rollout")
|
228 |
],
|
229 |
+
title="Attention Map Visualizer for timm Models",
|
|
|
230 |
description="Upload an image and select a timm model to visualize its attention maps."
|
231 |
)
|
232 |
|
233 |
+
iface.launch()
|