Spaces:
Runtime error
Runtime error
Commit
·
1129909
1
Parent(s):
424ddf2
Update app.py
Browse files
app.py
CHANGED
@@ -1,28 +1,30 @@
|
|
1 |
import gradio as gr
|
2 |
-
|
3 |
-
from matplotlib import gridspec
|
4 |
-
import matplotlib.pyplot as plt
|
5 |
import numpy as np
|
6 |
-
from PIL import Image
|
7 |
import tensorflow as tf
|
8 |
-
from
|
|
|
|
|
|
|
9 |
|
10 |
-
|
11 |
-
|
12 |
-
)
|
13 |
-
model = TFSegformerForSemanticSegmentation.from_pretrained(
|
14 |
-
"nvidia/segformer-b0-finetuned-cityscapes-1024-1024"
|
15 |
-
)
|
16 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
17 |
def ade_palette():
|
18 |
-
"""ADE20K palette that maps each class to RGB values."""
|
19 |
return [
|
20 |
[255, 0, 0],
|
21 |
[255, 187, 0],
|
22 |
[255, 228, 0],
|
23 |
[29, 219, 22],
|
24 |
[178, 204, 255],
|
25 |
-
[1, 0, 255],
|
26 |
[165, 102, 255],
|
27 |
[217, 65, 197],
|
28 |
[116, 116, 116],
|
@@ -37,30 +39,25 @@ def ade_palette():
|
|
37 |
[153, 0, 76]
|
38 |
]
|
39 |
|
40 |
-
labels_list = []
|
41 |
-
|
42 |
-
with open(r'labels.txt', 'r') as fp:
|
43 |
-
for line in fp:
|
44 |
-
labels_list.append(line[:-1])
|
45 |
-
|
46 |
colormap = np.asarray(ade_palette())
|
47 |
|
|
|
48 |
def label_to_color_image(label):
|
49 |
if label.ndim != 2:
|
50 |
raise ValueError("Expect 2-D input label")
|
51 |
-
|
52 |
if np.max(label) >= len(colormap):
|
53 |
raise ValueError("label value too large.")
|
54 |
return colormap[label]
|
55 |
|
|
|
56 |
def draw_plot(pred_img, seg):
|
57 |
fig = plt.figure(figsize=(20, 15))
|
58 |
-
|
59 |
grid_spec = gridspec.GridSpec(1, 2, width_ratios=[6, 1])
|
60 |
|
61 |
plt.subplot(grid_spec[0])
|
62 |
plt.imshow(pred_img)
|
63 |
plt.axis('off')
|
|
|
64 |
LABEL_NAMES = np.asarray(labels_list)
|
65 |
FULL_LABEL_MAP = np.arange(len(LABEL_NAMES)).reshape(len(LABEL_NAMES), 1)
|
66 |
FULL_COLOR_MAP = label_to_color_image(FULL_LABEL_MAP)
|
@@ -74,6 +71,7 @@ def draw_plot(pred_img, seg):
|
|
74 |
ax.tick_params(width=0.0, labelsize=25)
|
75 |
return fig
|
76 |
|
|
|
77 |
def sepia(input_img):
|
78 |
input_img = Image.fromarray(input_img)
|
79 |
|
@@ -84,27 +82,28 @@ def sepia(input_img):
|
|
84 |
logits = tf.transpose(logits, [0, 2, 3, 1])
|
85 |
logits = tf.image.resize(
|
86 |
logits, input_img.size[::-1]
|
87 |
-
)
|
88 |
-
seg = tf.math.argmax(logits, axis=-1)[0]
|
89 |
|
|
|
90 |
color_seg = np.zeros(
|
91 |
(seg.shape[0], seg.shape[1], 3), dtype=np.uint8
|
92 |
-
)
|
|
|
93 |
for label, color in enumerate(colormap):
|
94 |
color_seg[seg.numpy() == label, :] = color
|
95 |
|
96 |
-
# Show image + mask
|
97 |
pred_img = np.array(input_img) * 0.5 + color_seg * 0.5
|
98 |
pred_img = pred_img.astype(np.uint8)
|
99 |
|
100 |
fig = draw_plot(pred_img, seg)
|
101 |
return fig
|
102 |
|
|
|
103 |
demo = gr.Interface(fn=sepia,
|
104 |
inputs=gr.Image(shape=(800, 1200)),
|
105 |
outputs=['plot'],
|
106 |
examples=["citiscape-1.jpg", "citiscape-2.jpg"],
|
107 |
allow_flagging='never')
|
108 |
|
109 |
-
|
110 |
demo.launch()
|
|
|
1 |
import gradio as gr
|
|
|
|
|
|
|
2 |
import numpy as np
|
|
|
3 |
import tensorflow as tf
|
4 |
+
from PIL import Image
|
5 |
+
from transformers import SegformerImageProcessor, TFSegformerForSemanticSegmentation
|
6 |
+
import matplotlib.pyplot as plt
|
7 |
+
from matplotlib import gridspec
|
8 |
|
9 |
+
# Load model and feature extractor
|
10 |
+
feature_extractor = SegformerImageProcessor.from_pretrained("nvidia/segformer-b0-finetuned-cityscapes-1024-1024")
|
11 |
+
model = TFSegformerForSemanticSegmentation.from_pretrained("nvidia/segformer-b0-finetuned-cityscapes-1024-1024")
|
|
|
|
|
|
|
12 |
|
13 |
+
# Load labels
|
14 |
+
labels_list = []
|
15 |
+
with open(r'labels.txt', 'r') as fp:
|
16 |
+
for line in fp:
|
17 |
+
labels_list.append(line[:-1])
|
18 |
+
|
19 |
+
# ADE20K palette
|
20 |
def ade_palette():
|
|
|
21 |
return [
|
22 |
[255, 0, 0],
|
23 |
[255, 187, 0],
|
24 |
[255, 228, 0],
|
25 |
[29, 219, 22],
|
26 |
[178, 204, 255],
|
27 |
+
[1, 0, 255],
|
28 |
[165, 102, 255],
|
29 |
[217, 65, 197],
|
30 |
[116, 116, 116],
|
|
|
39 |
[153, 0, 76]
|
40 |
]
|
41 |
|
|
|
|
|
|
|
|
|
|
|
|
|
42 |
colormap = np.asarray(ade_palette())
|
43 |
|
44 |
+
# Label to color image mapping
|
45 |
def label_to_color_image(label):
|
46 |
if label.ndim != 2:
|
47 |
raise ValueError("Expect 2-D input label")
|
|
|
48 |
if np.max(label) >= len(colormap):
|
49 |
raise ValueError("label value too large.")
|
50 |
return colormap[label]
|
51 |
|
52 |
+
# Draw segmentation plot
|
53 |
def draw_plot(pred_img, seg):
|
54 |
fig = plt.figure(figsize=(20, 15))
|
|
|
55 |
grid_spec = gridspec.GridSpec(1, 2, width_ratios=[6, 1])
|
56 |
|
57 |
plt.subplot(grid_spec[0])
|
58 |
plt.imshow(pred_img)
|
59 |
plt.axis('off')
|
60 |
+
|
61 |
LABEL_NAMES = np.asarray(labels_list)
|
62 |
FULL_LABEL_MAP = np.arange(len(LABEL_NAMES)).reshape(len(LABEL_NAMES), 1)
|
63 |
FULL_COLOR_MAP = label_to_color_image(FULL_LABEL_MAP)
|
|
|
71 |
ax.tick_params(width=0.0, labelsize=25)
|
72 |
return fig
|
73 |
|
74 |
+
# Sepia function
|
75 |
def sepia(input_img):
|
76 |
input_img = Image.fromarray(input_img)
|
77 |
|
|
|
82 |
logits = tf.transpose(logits, [0, 2, 3, 1])
|
83 |
logits = tf.image.resize(
|
84 |
logits, input_img.size[::-1]
|
85 |
+
)
|
|
|
86 |
|
87 |
+
seg = tf.math.argmax(logits, axis=-1)[0]
|
88 |
color_seg = np.zeros(
|
89 |
(seg.shape[0], seg.shape[1], 3), dtype=np.uint8
|
90 |
+
)
|
91 |
+
|
92 |
for label, color in enumerate(colormap):
|
93 |
color_seg[seg.numpy() == label, :] = color
|
94 |
|
|
|
95 |
pred_img = np.array(input_img) * 0.5 + color_seg * 0.5
|
96 |
pred_img = pred_img.astype(np.uint8)
|
97 |
|
98 |
fig = draw_plot(pred_img, seg)
|
99 |
return fig
|
100 |
|
101 |
+
# Gradio Interface
|
102 |
demo = gr.Interface(fn=sepia,
|
103 |
inputs=gr.Image(shape=(800, 1200)),
|
104 |
outputs=['plot'],
|
105 |
examples=["citiscape-1.jpg", "citiscape-2.jpg"],
|
106 |
allow_flagging='never')
|
107 |
|
108 |
+
# Launch the interface
|
109 |
demo.launch()
|