Spaces:
Build error
Build error
Upload app.py
Browse files
app.py
ADDED
@@ -0,0 +1,130 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
import cv2
|
3 |
+
import sys
|
4 |
+
import numpy as np
|
5 |
+
import gradio as gr
|
6 |
+
from PIL import Image
|
7 |
+
import matplotlib.pyplot as plt
|
8 |
+
from segment_anything import sam_model_registry, SamAutomaticMaskGenerator
|
9 |
+
|
10 |
+
|
11 |
+
models = {
|
12 |
+
'vit_b': './checkpoints/sam_vit_b_01ec64.pth',
|
13 |
+
'vit_l': './checkpoints/sam_vit_l_0b3195.pth',
|
14 |
+
'vit_h': './checkpoints/sam_vit_h_4b8939.pth'
|
15 |
+
}
|
16 |
+
|
17 |
+
def inference(device, model_type, input_img, points_per_side, pred_iou_thresh, stability_score_thresh, min_mask_region_area,
|
18 |
+
stability_score_offset, box_nms_thresh, crop_n_layers, crop_nms_thresh):
|
19 |
+
sam = sam_model_registry[model_type](checkpoint=models[model_type]).to(device)
|
20 |
+
mask_generator = SamAutomaticMaskGenerator(
|
21 |
+
sam,
|
22 |
+
points_per_side=points_per_side,
|
23 |
+
pred_iou_thresh=pred_iou_thresh,
|
24 |
+
stability_score_thresh=stability_score_thresh,
|
25 |
+
stability_score_offset=stability_score_offset,
|
26 |
+
box_nms_thresh=box_nms_thresh,
|
27 |
+
crop_n_layers=crop_n_layers,
|
28 |
+
crop_nms_thresh=crop_nms_thresh,
|
29 |
+
crop_overlap_ratio=512 / 1500,
|
30 |
+
crop_n_points_downscale_factor=1,
|
31 |
+
point_grids=None,
|
32 |
+
min_mask_region_area=min_mask_region_area,
|
33 |
+
output_mode='binary_mask'
|
34 |
+
)
|
35 |
+
|
36 |
+
masks = mask_generator.generate(input_img)
|
37 |
+
sorted_anns = sorted(masks, key=(lambda x: x['area']), reverse=True)
|
38 |
+
|
39 |
+
mask_all = np.ones((input_img.shape[0], input_img.shape[1], 3))
|
40 |
+
for ann in sorted_anns:
|
41 |
+
m = ann['segmentation']
|
42 |
+
color_mask = np.random.random((1, 3)).tolist()[0]
|
43 |
+
for i in range(3):
|
44 |
+
mask_all[m==True, i] = color_mask[i]
|
45 |
+
result = input_img / 255 * 0.3 + mask_all * 0.7
|
46 |
+
|
47 |
+
return result, mask_all
|
48 |
+
|
49 |
+
|
50 |
+
|
51 |
+
with gr.Blocks() as demo:
|
52 |
+
with gr.Row():
|
53 |
+
gr.Markdown(
|
54 |
+
'''# Segment Anything!🚀
|
55 |
+
分割一切!CV的GPT-3时刻!
|
56 |
+
[**官方网址**](https://segment-anything.com/)
|
57 |
+
'''
|
58 |
+
)
|
59 |
+
with gr.Row():
|
60 |
+
# 选择模型类型
|
61 |
+
model_type = gr.Dropdown(["vit_b", "vit_l", "vit_h"], value='vit_b', label="选择模型")
|
62 |
+
# 选择device
|
63 |
+
device = gr.Dropdown(["cpu", "cuda"], value='cuda', label="选择你的硬件")
|
64 |
+
|
65 |
+
# 参数
|
66 |
+
with gr.Accordion(label='参数调整', open=False):
|
67 |
+
with gr.Row():
|
68 |
+
points_per_side = gr.Number(value=32, label="points_per_side", precision=0,
|
69 |
+
info='''The number of points to be sampled along one side of the image. The total
|
70 |
+
number of points is points_per_side**2.''')
|
71 |
+
pred_iou_thresh = gr.Slider(value=0.88, minimum=0, maximum=1.0, step=0.01, label="pred_iou_thresh",
|
72 |
+
info='''A filtering threshold in [0,1], using the model's predicted mask quality.''')
|
73 |
+
stability_score_thresh = gr.Slider(value=0.95, minimum=0, maximum=1.0, step=0.01, label="stability_score_thresh",
|
74 |
+
info='''A filtering threshold in [0,1], using the stability of the mask under
|
75 |
+
changes to the cutoff used to binarize the model's mask predictions.''')
|
76 |
+
min_mask_region_area = gr.Number(value=0, label="min_mask_region_area", precision=0,
|
77 |
+
info='''If >0, postprocessing will be applied to remove disconnected regions
|
78 |
+
and holes in masks with area smaller than min_mask_region_area.''')
|
79 |
+
with gr.Row():
|
80 |
+
stability_score_offset = gr.Number(value=1, label="stability_score_offset",
|
81 |
+
info='''The amount to shift the cutoff when calculated the stability score.''')
|
82 |
+
box_nms_thresh = gr.Slider(value=0.7, minimum=0, maximum=1.0, step=0.01, label="box_nms_thresh",
|
83 |
+
info='''The box IoU cutoff used by non-maximal ression to filter duplicate masks.''')
|
84 |
+
crop_n_layers = gr.Number(value=0, label="crop_n_layers", precision=0,
|
85 |
+
info='''If >0, mask prediction will be run again on crops of the image.
|
86 |
+
Sets the number of layers to run, where each layer has 2**i_layer number of image crops.''')
|
87 |
+
crop_nms_thresh = gr.Slider(value=0.7, minimum=0, maximum=1.0, step=0.01, label="crop_nms_thresh",
|
88 |
+
info='''The box IoU cutoff used by non-maximal suppression to filter duplicate
|
89 |
+
masks between different crops.''')
|
90 |
+
|
91 |
+
# 显示图片
|
92 |
+
with gr.Row().style(equal_height=True):
|
93 |
+
with gr.Column():
|
94 |
+
input_image = gr.Image(type="numpy")
|
95 |
+
with gr.Row():
|
96 |
+
button = gr.Button("Auto!")
|
97 |
+
with gr.Tab(label='原图+mask'):
|
98 |
+
image_output = gr.Image(type='numpy')
|
99 |
+
with gr.Tab(label='Mask'):
|
100 |
+
mask_output = gr.Image(type='numpy')
|
101 |
+
|
102 |
+
gr.Examples(
|
103 |
+
examples=[os.path.join(os.path.dirname(__file__), "./images/53960-scaled.jpg"),
|
104 |
+
os.path.join(os.path.dirname(__file__), "./images/2388455-scaled.jpg"),
|
105 |
+
os.path.join(os.path.dirname(__file__), "./images/1.jpg"),
|
106 |
+
os.path.join(os.path.dirname(__file__), "./images/2.jpg"),
|
107 |
+
os.path.join(os.path.dirname(__file__), "./images/3.jpg"),
|
108 |
+
os.path.join(os.path.dirname(__file__), "./images/4.jpg"),
|
109 |
+
os.path.join(os.path.dirname(__file__), "./images/5.jpg"),
|
110 |
+
os.path.join(os.path.dirname(__file__), "./images/6.jpg"),
|
111 |
+
os.path.join(os.path.dirname(__file__), "./images/7.jpg"),
|
112 |
+
os.path.join(os.path.dirname(__file__), "./images/8.jpg"),
|
113 |
+
],
|
114 |
+
inputs=input_image,
|
115 |
+
outputs=image_output,
|
116 |
+
)
|
117 |
+
|
118 |
+
|
119 |
+
# 按钮交互
|
120 |
+
button.click(inference, inputs=[device, model_type, input_image, points_per_side, pred_iou_thresh,
|
121 |
+
stability_score_thresh, min_mask_region_area, stability_score_offset, box_nms_thresh,
|
122 |
+
crop_n_layers, crop_nms_thresh],
|
123 |
+
outputs=[image_output, mask_output])
|
124 |
+
|
125 |
+
|
126 |
+
|
127 |
+
demo.launch(debug=True)
|
128 |
+
|
129 |
+
|
130 |
+
|