Spaces:
Sleeping
Sleeping
Update
Browse files- README.md +1 -1
- app.py +32 -13
- requirements.txt +3 -2
README.md
CHANGED
@@ -4,7 +4,7 @@ emoji: 🏃
|
|
4 |
colorFrom: gray
|
5 |
colorTo: purple
|
6 |
sdk: gradio
|
7 |
-
sdk_version: 3.
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
---
|
|
|
4 |
colorFrom: gray
|
5 |
colorTo: purple
|
6 |
sdk: gradio
|
7 |
+
sdk_version: 3.37.0
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
---
|
app.py
CHANGED
@@ -47,8 +47,9 @@ model = load_model()
|
|
47 |
labels = load_labels()
|
48 |
|
49 |
|
50 |
-
def predict(
|
51 |
-
|
|
|
52 |
_, height, width, _ = model.input_shape
|
53 |
image = np.asarray(image)
|
54 |
image = tf.image.resize(image,
|
@@ -60,12 +61,19 @@ def predict(image: PIL.Image.Image,
|
|
60 |
image = image / 255.
|
61 |
probs = model.predict(image[None, ...])[0]
|
62 |
probs = probs.astype(float)
|
63 |
-
|
64 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
65 |
if prob < score_threshold:
|
66 |
-
|
67 |
-
|
68 |
-
|
|
|
69 |
|
70 |
|
71 |
image_paths = load_sample_image_paths()
|
@@ -83,15 +91,26 @@ with gr.Blocks(css='style.css') as demo:
|
|
83 |
value=0.5)
|
84 |
run_button = gr.Button('Run')
|
85 |
with gr.Column():
|
86 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
87 |
gr.Examples(examples=examples,
|
88 |
inputs=[image, score_threshold],
|
89 |
-
outputs=result,
|
90 |
fn=predict,
|
91 |
cache_examples=os.getenv('CACHE_EXAMPLES') == '1')
|
92 |
|
93 |
-
run_button.click(
|
94 |
-
|
95 |
-
|
96 |
-
|
|
|
|
|
97 |
demo.queue().launch()
|
|
|
47 |
labels = load_labels()
|
48 |
|
49 |
|
50 |
+
def predict(
|
51 |
+
image: PIL.Image.Image, score_threshold: float
|
52 |
+
) -> tuple[dict[str, float], dict[str, float], str]:
|
53 |
_, height, width, _ = model.input_shape
|
54 |
image = np.asarray(image)
|
55 |
image = tf.image.resize(image,
|
|
|
61 |
image = image / 255.
|
62 |
probs = model.predict(image[None, ...])[0]
|
63 |
probs = probs.astype(float)
|
64 |
+
|
65 |
+
indices = np.argsort(probs)[::-1]
|
66 |
+
result_all = dict()
|
67 |
+
result_threshold = dict()
|
68 |
+
for index in indices:
|
69 |
+
label = labels[index]
|
70 |
+
prob = probs[index]
|
71 |
+
result_all[label] = prob
|
72 |
if prob < score_threshold:
|
73 |
+
break
|
74 |
+
result_threshold[label] = prob
|
75 |
+
result_text = ', '.join(result_all.keys())
|
76 |
+
return result_threshold, result_all, result_text
|
77 |
|
78 |
|
79 |
image_paths = load_sample_image_paths()
|
|
|
91 |
value=0.5)
|
92 |
run_button = gr.Button('Run')
|
93 |
with gr.Column():
|
94 |
+
with gr.Tabs():
|
95 |
+
with gr.Tab(label='Output'):
|
96 |
+
result = gr.Label(label='Output', show_label=False)
|
97 |
+
with gr.Tab(label='JSON'):
|
98 |
+
result_json = gr.JSON(label='JSON output',
|
99 |
+
show_label=False)
|
100 |
+
with gr.Tab(label='Text'):
|
101 |
+
result_text = gr.Text(label='Text output',
|
102 |
+
show_label=False,
|
103 |
+
lines=5)
|
104 |
gr.Examples(examples=examples,
|
105 |
inputs=[image, score_threshold],
|
106 |
+
outputs=[result, result_json, result_text],
|
107 |
fn=predict,
|
108 |
cache_examples=os.getenv('CACHE_EXAMPLES') == '1')
|
109 |
|
110 |
+
run_button.click(
|
111 |
+
fn=predict,
|
112 |
+
inputs=[image, score_threshold],
|
113 |
+
outputs=[result, result_json, result_text],
|
114 |
+
api_name='predict',
|
115 |
+
)
|
116 |
demo.queue().launch()
|
requirements.txt
CHANGED
@@ -1,3 +1,4 @@
|
|
1 |
-
pillow>=9.0.0
|
2 |
-
tensorflow>=2.7.0
|
3 |
git+https://github.com/KichangKim/DeepDanbooru@v3-20200915-sgd-e30#egg=deepdanbooru
|
|
|
|
|
|
|
|
|
|
|
|
1 |
git+https://github.com/KichangKim/DeepDanbooru@v3-20200915-sgd-e30#egg=deepdanbooru
|
2 |
+
pillow==10.0.0
|
3 |
+
pydantic==1.10.11
|
4 |
+
tensorflow==2.13.0
|