Spaces:
Build error
Build error
sagemaker
commited on
Commit
·
7b843a4
1
Parent(s):
bc954b8
some upgrades
Browse files
app.py
CHANGED
@@ -4,13 +4,21 @@ from PIL import Image
|
|
4 |
import torch
|
5 |
import matplotlib.pyplot as plt
|
6 |
import io
|
|
|
7 |
|
8 |
|
9 |
COLORS = [[0.000, 0.447, 0.741], [0.850, 0.325, 0.098], [0.929, 0.694, 0.125],
|
10 |
[0.494, 0.184, 0.556], [0.466, 0.674, 0.188], [0.301, 0.745, 0.933]]
|
11 |
|
12 |
|
13 |
-
def
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
14 |
feature_extractor = AutoFeatureExtractor.from_pretrained(f"hustvl/{model_name}")
|
15 |
model = YolosForObjectDetection.from_pretrained(f"hustvl/{model_name}")
|
16 |
|
@@ -22,26 +30,33 @@ def infer(img, model_name):
|
|
22 |
outputs = model(pixel_values, output_attentions=True)
|
23 |
|
24 |
probas = outputs.logits.softmax(-1)[0, :, :-1]
|
25 |
-
keep = probas.max(-1).values >
|
26 |
|
27 |
target_sizes = torch.tensor(img.size[::-1]).unsqueeze(0)
|
28 |
postprocessed_outputs = feature_extractor.post_process(outputs, target_sizes)
|
29 |
bboxes_scaled = postprocessed_outputs[0]['boxes']
|
30 |
|
31 |
-
|
|
|
32 |
|
33 |
return res_img
|
34 |
|
35 |
-
def plot_results(pil_img, prob, boxes, model):
|
36 |
plt.figure(figsize=(16,10))
|
37 |
plt.imshow(pil_img)
|
38 |
ax = plt.gca()
|
39 |
colors = COLORS * 100
|
40 |
for p, (xmin, ymin, xmax, ymax), c in zip(prob, boxes.tolist(), colors):
|
41 |
-
ax.add_patch(plt.Rectangle((xmin, ymin), xmax - xmin, ymax - ymin,
|
42 |
-
fill=False, color=c, linewidth=3))
|
43 |
cl = p.argmax()
|
44 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
45 |
ax.text(xmin, ymin, text, fontsize=15,
|
46 |
bbox=dict(facecolor='yellow', alpha=0.5))
|
47 |
plt.axis('off')
|
@@ -54,17 +69,25 @@ def fig2img(fig):
|
|
54 |
img = Image.open(buf)
|
55 |
return img
|
56 |
|
57 |
-
description = """Object Detection with YOLOS. Choose
|
|
|
|
|
|
|
|
|
|
|
|
|
58 |
|
59 |
image_in = gr.components.Image()
|
60 |
image_out = gr.components.Image()
|
61 |
model_choice = gr.components.Dropdown(["yolos-tiny", "yolos-small", "yolos_base", "yolos-small-300", "yolos-small-dwr"], value="yolos-small")
|
|
|
|
|
62 |
|
63 |
Iface = gr.Interface(
|
64 |
fn=infer,
|
65 |
-
inputs=[image_in,model_choice],
|
66 |
outputs=image_out,
|
67 |
-
examples=[["examples/10_People_Marching_People_Marching_2_120.jpg"], ["examples/12_Group_Group_12_Group_Group_12_26.jpg"], ["examples/43_Row_Boat_Canoe_43_247.jpg"]],
|
68 |
title="Object Detection with YOLOS",
|
69 |
description=description,
|
70 |
).launch()
|
|
|
4 |
import torch
|
5 |
import matplotlib.pyplot as plt
|
6 |
import io
|
7 |
+
import numpy as np
|
8 |
|
9 |
|
10 |
COLORS = [[0.000, 0.447, 0.741], [0.850, 0.325, 0.098], [0.929, 0.694, 0.125],
|
11 |
[0.494, 0.184, 0.556], [0.466, 0.674, 0.188], [0.301, 0.745, 0.933]]
|
12 |
|
13 |
|
14 |
+
def get_class_list_from_input(classes_string: str):
|
15 |
+
if classes_string == "":
|
16 |
+
return []
|
17 |
+
classes_list = classes_string.split(",")
|
18 |
+
classes_list = [x.strip() for x in classes_list]
|
19 |
+
return classes_list
|
20 |
+
|
21 |
+
def infer(img, model_name: str, prob_threshold: int, classes_to_show = str):
|
22 |
feature_extractor = AutoFeatureExtractor.from_pretrained(f"hustvl/{model_name}")
|
23 |
model = YolosForObjectDetection.from_pretrained(f"hustvl/{model_name}")
|
24 |
|
|
|
30 |
outputs = model(pixel_values, output_attentions=True)
|
31 |
|
32 |
probas = outputs.logits.softmax(-1)[0, :, :-1]
|
33 |
+
keep = probas.max(-1).values > prob_threshold
|
34 |
|
35 |
target_sizes = torch.tensor(img.size[::-1]).unsqueeze(0)
|
36 |
postprocessed_outputs = feature_extractor.post_process(outputs, target_sizes)
|
37 |
bboxes_scaled = postprocessed_outputs[0]['boxes']
|
38 |
|
39 |
+
classes_list = get_class_list_from_input(classes_to_show)
|
40 |
+
res_img = plot_results(img, probas[keep], bboxes_scaled[keep], model, classes_list)
|
41 |
|
42 |
return res_img
|
43 |
|
44 |
+
def plot_results(pil_img, prob, boxes, model, classes_list):
|
45 |
plt.figure(figsize=(16,10))
|
46 |
plt.imshow(pil_img)
|
47 |
ax = plt.gca()
|
48 |
colors = COLORS * 100
|
49 |
for p, (xmin, ymin, xmax, ymax), c in zip(prob, boxes.tolist(), colors):
|
|
|
|
|
50 |
cl = p.argmax()
|
51 |
+
object_class = model.config.id2label[cl.item()]
|
52 |
+
|
53 |
+
if len(classes_list) > 0 :
|
54 |
+
if object_class not in classes_list:
|
55 |
+
continue
|
56 |
+
|
57 |
+
ax.add_patch(plt.Rectangle((xmin, ymin), xmax - xmin, ymax - ymin,
|
58 |
+
fill=False, color=c, linewidth=3))
|
59 |
+
text = f'{object_class}: {p[cl]:0.2f}'
|
60 |
ax.text(xmin, ymin, text, fontsize=15,
|
61 |
bbox=dict(facecolor='yellow', alpha=0.5))
|
62 |
plt.axis('off')
|
|
|
69 |
img = Image.open(buf)
|
70 |
return img
|
71 |
|
72 |
+
description = """Object Detection with YOLOS. Choose https://github.com/amikelive/coco-labels/blob/master/coco-labels-2014_2017.txtyour model and you're good to go.
|
73 |
+
|
74 |
+
You can adapt the minimum probability threshold with the slider.
|
75 |
+
|
76 |
+
Additionally you can restrict the classes that will be shown by putting in a comma separated list of
|
77 |
+
[COCO classes](https://github.com/amikelive/coco-labels/blob/master/coco-labels-2014_2017.txt).
|
78 |
+
Leaving the field empty will show all classes"""
|
79 |
|
80 |
image_in = gr.components.Image()
|
81 |
image_out = gr.components.Image()
|
82 |
model_choice = gr.components.Dropdown(["yolos-tiny", "yolos-small", "yolos_base", "yolos-small-300", "yolos-small-dwr"], value="yolos-small")
|
83 |
+
prob_threshold_slider = gr.components.Slider(minimum=0, maximum=1.0, step=0.01, value=0.9, label="Probability Threshold")
|
84 |
+
classes_to_show = gr.components.Textbox(placeholder="e.g. person, boat")
|
85 |
|
86 |
Iface = gr.Interface(
|
87 |
fn=infer,
|
88 |
+
inputs=[image_in,model_choice, prob_threshold_slider, classes_to_show],
|
89 |
outputs=image_out,
|
90 |
+
#examples=[["examples/10_People_Marching_People_Marching_2_120.jpg"], ["examples/12_Group_Group_12_Group_Group_12_26.jpg"], ["examples/43_Row_Boat_Canoe_43_247.jpg"]],
|
91 |
title="Object Detection with YOLOS",
|
92 |
description=description,
|
93 |
).launch()
|