Spaces:
Sleeping
Sleeping
Yuxiang Wang
commited on
Commit
·
c5343e6
1
Parent(s):
af9c1e6
explanations,closest sample
Browse files- app.py +64 -22
- closest_sample.py +16 -9
- explanations.py +8 -3
- inference_beit.py +203 -0
app.py
CHANGED
@@ -46,9 +46,19 @@ def get_model(model_name):
|
|
46 |
nb_classes = n_classes,load_weights=False,finer_model=True,backbone_name ='Resnet50v2')
|
47 |
model.load_weights('model_classification/rock-170.h5')
|
48 |
else:
|
49 |
-
|
50 |
return model,n_classes
|
51 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
52 |
def segment_image(input_image):
|
53 |
img = segmentation_sam(input_image)
|
54 |
return img
|
@@ -67,7 +77,8 @@ def classify_image(input_image, model_name):
|
|
67 |
if 'Fossils 19' ==model_name:
|
68 |
from inference_beit import inference_dino
|
69 |
model,n_classes = get_model(model_name)
|
70 |
-
|
|
|
71 |
return None
|
72 |
|
73 |
def get_embeddings(input_image,model_name):
|
@@ -84,21 +95,26 @@ def get_embeddings(input_image,model_name):
|
|
84 |
if 'Fossils 19' ==model_name:
|
85 |
from inference_beit import inference_dino
|
86 |
model,n_classes = get_model(model_name)
|
87 |
-
|
|
|
|
|
|
|
88 |
return None
|
89 |
|
90 |
|
91 |
def find_closest(input_image,model_name):
|
92 |
embedding = get_embeddings(input_image,model_name)
|
93 |
-
paths = get_images(embedding)
|
94 |
-
|
|
|
95 |
|
96 |
def explain_image(input_image,model_name):
|
97 |
model,n_classes= get_model(model_name)
|
98 |
-
saliency, integrated, smoothgrad
|
|
|
99 |
#original = saliency + integrated + smoothgrad
|
100 |
print('done')
|
101 |
-
return
|
102 |
|
103 |
#minimalist theme
|
104 |
with gr.Blocks(theme='sudeepshouche/minimalist') as demo:
|
@@ -118,7 +134,7 @@ with gr.Blocks(theme='sudeepshouche/minimalist') as demo:
|
|
118 |
|
119 |
with gr.Column():
|
120 |
model_name = gr.Dropdown(
|
121 |
-
["Mummified 170", "Rock 170"],
|
122 |
multiselect=False,
|
123 |
value="Rock 170", # default option
|
124 |
label="Model",
|
@@ -142,32 +158,61 @@ with gr.Blocks(theme='sudeepshouche/minimalist') as demo:
|
|
142 |
# with gr.Column():
|
143 |
# class_predicted2 = gr.Label(label='Class Predicted from diffuser')
|
144 |
# classify_button = gr.Button("Classify Image")
|
145 |
-
|
146 |
|
147 |
with gr.Accordion("Explanations "):
|
148 |
gr.Markdown("Computing Explanations from the model")
|
149 |
with gr.Row():
|
150 |
#original_input = gr.Image(label="Original Frame")
|
151 |
-
saliency = gr.Image(label="saliency")
|
152 |
-
gradcam = gr.Image(label='integraged gradients')
|
153 |
-
guided_gradcam = gr.Image(label='gradcam')
|
154 |
#guided_backprop = gr.Image(label='guided backprop')
|
|
|
155 |
generate_explanations = gr.Button("Generate Explanations")
|
156 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
157 |
with gr.Accordion('Closest Images'):
|
158 |
gr.Markdown("Finding the closest images in the dataset")
|
|
|
159 |
with gr.Row():
|
160 |
-
|
161 |
-
|
162 |
-
|
163 |
-
closest_image_3 = gr.Image(label='Forth Closest Image')
|
164 |
-
closest_image_4 = gr.Image(label='Fifth Closest Image')
|
165 |
find_closest_btn = gr.Button("Find Closest Images")
|
166 |
|
167 |
segment_button.click(segment_image, inputs=input_image, outputs=segmented_image)
|
168 |
classify_image_button.click(classify_image, inputs=[input_image,model_name], outputs=class_predicted)
|
169 |
-
generate_explanations.click(explain_image, inputs=[input_image,model_name], outputs=[saliency,gradcam,guided_gradcam
|
170 |
-
find_closest_btn.click(find_closest, inputs=[input_image,model_name], outputs=[closest_image_0,closest_image_1,closest_image_2,closest_image_3,closest_image_4])
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
171 |
#classify_segmented_button.click(classify_image, inputs=[segmented_image,model_name], outputs=class_predicted)
|
172 |
|
173 |
demo.queue() # manage multiple incoming requests
|
@@ -176,6 +221,3 @@ if os.getenv('SYSTEM') == 'spaces':
|
|
176 |
demo.launch(width='40%',auth=(os.environ.get('USERNAME'), os.environ.get('PASSWORD')))
|
177 |
else:
|
178 |
demo.launch()
|
179 |
-
|
180 |
-
|
181 |
-
|
|
|
46 |
nb_classes = n_classes,load_weights=False,finer_model=True,backbone_name ='Resnet50v2')
|
47 |
model.load_weights('model_classification/rock-170.h5')
|
48 |
else:
|
49 |
+
raise ValueError(f"Model name '{model_name}' is not recognized")
|
50 |
return model,n_classes
|
51 |
|
52 |
+
'''
|
53 |
+
elif model_name == 'Fossils 19':
|
54 |
+
n_classes = 19 or 23?
|
55 |
+
model = get_beit_model(input_shape=(600, 600, 3),
|
56 |
+
num_labels=n_classes,
|
57 |
+
load_weights=False,
|
58 |
+
)
|
59 |
+
model.load_weights('model_classification/beit-fossils-19.h5')
|
60 |
+
'''
|
61 |
+
|
62 |
def segment_image(input_image):
|
63 |
img = segmentation_sam(input_image)
|
64 |
return img
|
|
|
77 |
if 'Fossils 19' ==model_name:
|
78 |
from inference_beit import inference_dino
|
79 |
model,n_classes = get_model(model_name)
|
80 |
+
result = inference_dino(input_image,model_name)
|
81 |
+
return result
|
82 |
return None
|
83 |
|
84 |
def get_embeddings(input_image,model_name):
|
|
|
95 |
if 'Fossils 19' ==model_name:
|
96 |
from inference_beit import inference_dino
|
97 |
model,n_classes = get_model(model_name)
|
98 |
+
result = inference_dino(input_image,model_name)
|
99 |
+
#TODO
|
100 |
+
#result = inference_beit_embedding
|
101 |
+
return result
|
102 |
return None
|
103 |
|
104 |
|
105 |
def find_closest(input_image,model_name):
|
106 |
embedding = get_embeddings(input_image,model_name)
|
107 |
+
classes, paths = get_images(embedding)
|
108 |
+
#outputs = classes+paths
|
109 |
+
return classes,paths
|
110 |
|
111 |
def explain_image(input_image,model_name):
|
112 |
model,n_classes= get_model(model_name)
|
113 |
+
#saliency, integrated, smoothgrad,
|
114 |
+
rise = explain(model,input_image,n_classes=n_classes)
|
115 |
#original = saliency + integrated + smoothgrad
|
116 |
print('done')
|
117 |
+
return rise
|
118 |
|
119 |
#minimalist theme
|
120 |
with gr.Blocks(theme='sudeepshouche/minimalist') as demo:
|
|
|
134 |
|
135 |
with gr.Column():
|
136 |
model_name = gr.Dropdown(
|
137 |
+
["Mummified 170", "Rock 170","Fossils 19"],
|
138 |
multiselect=False,
|
139 |
value="Rock 170", # default option
|
140 |
label="Model",
|
|
|
158 |
# with gr.Column():
|
159 |
# class_predicted2 = gr.Label(label='Class Predicted from diffuser')
|
160 |
# classify_button = gr.Button("Classify Image")
|
161 |
+
|
162 |
|
163 |
with gr.Accordion("Explanations "):
|
164 |
gr.Markdown("Computing Explanations from the model")
|
165 |
with gr.Row():
|
166 |
#original_input = gr.Image(label="Original Frame")
|
167 |
+
#saliency = gr.Image(label="saliency")
|
168 |
+
#gradcam = gr.Image(label='integraged gradients')
|
169 |
+
#guided_gradcam = gr.Image(label='gradcam')
|
170 |
#guided_backprop = gr.Image(label='guided backprop')
|
171 |
+
rise = gr.Image(label = 'Rise')
|
172 |
generate_explanations = gr.Button("Generate Explanations")
|
173 |
|
174 |
+
# with gr.Accordion('Closest Images'):
|
175 |
+
# gr.Markdown("Finding the closest images in the dataset")
|
176 |
+
# with gr.Row():
|
177 |
+
# with gr.Column():
|
178 |
+
# label_closest_image_0 = gr.Markdown('')
|
179 |
+
# closest_image_0 = gr.Image(label='Closest Image',image_mode='contain',width=200, height=200)
|
180 |
+
# with gr.Column():
|
181 |
+
# label_closest_image_1 = gr.Markdown('')
|
182 |
+
# closest_image_1 = gr.Image(label='Second Closest Image',image_mode='contain',width=200, height=200)
|
183 |
+
# with gr.Column():
|
184 |
+
# label_closest_image_2 = gr.Markdown('')
|
185 |
+
# closest_image_2 = gr.Image(label='Third Closest Image',image_mode='contain',width=200, height=200)
|
186 |
+
# with gr.Column():
|
187 |
+
# label_closest_image_3 = gr.Markdown('')
|
188 |
+
# closest_image_3 = gr.Image(label='Forth Closest Image',image_mode='contain', width=200, height=200)
|
189 |
+
# with gr.Column():
|
190 |
+
# label_closest_image_4 = gr.Markdown('')
|
191 |
+
# closest_image_4 = gr.Image(label='Fifth Closest Image',image_mode='contain',width=200, height=200)
|
192 |
+
# find_closest_btn = gr.Button("Find Closest Images")
|
193 |
with gr.Accordion('Closest Images'):
|
194 |
gr.Markdown("Finding the closest images in the dataset")
|
195 |
+
|
196 |
with gr.Row():
|
197 |
+
gallery = gr.Gallery(label="Closest Images", show_label=False,elem_id="gallery",columns=[5], rows=[1],height='auto', allow_preview=True, preview=None)
|
198 |
+
#.style(grid=[1, 5], height=200, width=200)
|
199 |
+
|
|
|
|
|
200 |
find_closest_btn = gr.Button("Find Closest Images")
|
201 |
|
202 |
segment_button.click(segment_image, inputs=input_image, outputs=segmented_image)
|
203 |
classify_image_button.click(classify_image, inputs=[input_image,model_name], outputs=class_predicted)
|
204 |
+
generate_explanations.click(explain_image, inputs=[input_image,model_name], outputs=[rise]) #saliency,gradcam,guided_gradcam
|
205 |
+
#find_closest_btn.click(find_closest, inputs=[input_image,model_name], outputs=[label_closest_image_0,label_closest_image_1,label_closest_image_2,label_closest_image_3,label_closest_image_4,closest_image_0,closest_image_1,closest_image_2,closest_image_3,closest_image_4])
|
206 |
+
def update_outputs(input_image,model_name):
|
207 |
+
labels, images = find_closest(input_image,model_name)
|
208 |
+
#labels_html = "".join([f'<div style="display: inline-block; text-align: center; width: 18%;">{label}</div>' for label in labels])
|
209 |
+
#labels_markdown = f"<div style='width: 100%; text-align: center;'>{labels_html}</div>"
|
210 |
+
image_caption=[]
|
211 |
+
for i in range(5):
|
212 |
+
image_caption.append((images[i],labels[i]))
|
213 |
+
return image_caption
|
214 |
+
|
215 |
+
find_closest_btn.click(fn=update_outputs, inputs=[input_image,model_name], outputs=[gallery])
|
216 |
#classify_segmented_button.click(classify_image, inputs=[segmented_image,model_name], outputs=class_predicted)
|
217 |
|
218 |
demo.queue() # manage multiple incoming requests
|
|
|
221 |
demo.launch(width='40%',auth=(os.environ.get('USERNAME'), os.environ.get('PASSWORD')))
|
222 |
else:
|
223 |
demo.launch()
|
|
|
|
|
|
closest_sample.py
CHANGED
@@ -50,10 +50,8 @@ def download_public_image(url, destination_path):
|
|
50 |
with open(destination_path, 'wb') as f:
|
51 |
f.write(response.content)
|
52 |
print(f"Downloaded image to {destination_path}")
|
53 |
-
return True
|
54 |
else:
|
55 |
print(f"Failed to download image from bucket. Status code: {response.status_code}")
|
56 |
-
return False
|
57 |
|
58 |
def get_images(embedding):
|
59 |
|
@@ -69,14 +67,23 @@ def get_images(embedding):
|
|
69 |
folder_florissant = 'https://storage.googleapis.com/serrelab/prj_fossils/2024/Florissant_Fossil_v2.0/'
|
70 |
folder_general = 'https://storage.googleapis.com/serrelab/prj_fossils/2024/General_Fossil_v2.0/'
|
71 |
|
|
|
|
|
72 |
for i, path in enumerate(paths):
|
73 |
local_file_path = f'image_{i}.jpg'
|
74 |
-
|
75 |
-
|
76 |
-
|
77 |
-
|
78 |
-
|
79 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
80 |
#paths= [path.replace('/gpfs/data/tserre/irodri15/Fossils/new_data/leavesdb-v1_1/images/Fossil/Florissant_Fossil/512/full/jpg/',
|
81 |
# '/media/data_cifs/projects/prj_fossils/data/processed_data/leavesdb-v1_1/images/Fossil/Florissant_Fossil/original/full/jpg/') for path in paths]
|
82 |
-
|
|
|
|
50 |
with open(destination_path, 'wb') as f:
|
51 |
f.write(response.content)
|
52 |
print(f"Downloaded image to {destination_path}")
|
|
|
53 |
else:
|
54 |
print(f"Failed to download image from bucket. Status code: {response.status_code}")
|
|
|
55 |
|
56 |
def get_images(embedding):
|
57 |
|
|
|
67 |
folder_florissant = 'https://storage.googleapis.com/serrelab/prj_fossils/2024/Florissant_Fossil_v2.0/'
|
68 |
folder_general = 'https://storage.googleapis.com/serrelab/prj_fossils/2024/General_Fossil_v2.0/'
|
69 |
|
70 |
+
local_paths = []
|
71 |
+
classes = []
|
72 |
for i, path in enumerate(paths):
|
73 |
local_file_path = f'image_{i}.jpg'
|
74 |
+
if 'Florissant_Fossil/512/full/jpg/' in path:
|
75 |
+
public_path = path.replace('/gpfs/data/tserre/irodri15/Fossils/new_data/leavesdb-v1_1/images/Fossil/Florissant_Fossil/512/full/jpg/', folder_florissant)
|
76 |
+
elif 'General_Fossil/512/full/jpg/' in path:
|
77 |
+
public_path = path.replace('/gpfs/data/tserre/irodri15/Fossils/new_data/leavesdb-v1_1/images/Fossil/General_Fossil/512/full/jpg/', folder_general)
|
78 |
+
else:
|
79 |
+
print("no match found")
|
80 |
+
download_public_image(public_path, local_file_path)
|
81 |
+
names = []
|
82 |
+
parts = [part for part in public_path.split('/') if part]
|
83 |
+
part = parts[-2]
|
84 |
+
classes.append(part)
|
85 |
+
local_paths.append(local_file_path)
|
86 |
#paths= [path.replace('/gpfs/data/tserre/irodri15/Fossils/new_data/leavesdb-v1_1/images/Fossil/Florissant_Fossil/512/full/jpg/',
|
87 |
# '/media/data_cifs/projects/prj_fossils/data/processed_data/leavesdb-v1_1/images/Fossil/Florissant_Fossil/original/full/jpg/') for path in paths]
|
88 |
+
|
89 |
+
return classes, local_paths
|
explanations.py
CHANGED
@@ -50,10 +50,13 @@ def explain(model, input_image,size=600, n_classes=171) :
|
|
50 |
class_model = tf.keras.Model(model.input, model.output[1])
|
51 |
|
52 |
explainers = [
|
53 |
-
Saliency(class_model),
|
54 |
-
IntegratedGradients(class_model, steps=50, batch_size=BATCH_SIZE),
|
55 |
-
SmoothGrad(class_model, nb_samples=50, batch_size=BATCH_SIZE),
|
56 |
#GradCAM(class_model),
|
|
|
|
|
|
|
57 |
]
|
58 |
cropped,repetitions = _clever_crop(input_image,(size,size))
|
59 |
size_repetitions = int(size//(repetitions.numpy()+1))
|
@@ -74,6 +77,8 @@ def explain(model, input_image,size=600, n_classes=171) :
|
|
74 |
|
75 |
plt.savefig(f'phi_{e}.png')
|
76 |
explanations.append(f'phi_{e}.png')
|
|
|
|
|
77 |
|
78 |
print('Done')
|
79 |
|
|
|
50 |
class_model = tf.keras.Model(model.input, model.output[1])
|
51 |
|
52 |
explainers = [
|
53 |
+
#Saliency(class_model),
|
54 |
+
#IntegratedGradients(class_model, steps=50, batch_size=BATCH_SIZE),
|
55 |
+
#SmoothGrad(class_model, nb_samples=50, batch_size=BATCH_SIZE),
|
56 |
#GradCAM(class_model),
|
57 |
+
Rise(class_model,nb_samples = 50, batch_size = BATCH_SIZE,grid_size=7,
|
58 |
+
preservation_probability=0.5)
|
59 |
+
#
|
60 |
]
|
61 |
cropped,repetitions = _clever_crop(input_image,(size,size))
|
62 |
size_repetitions = int(size//(repetitions.numpy()+1))
|
|
|
77 |
|
78 |
plt.savefig(f'phi_{e}.png')
|
79 |
explanations.append(f'phi_{e}.png')
|
80 |
+
print(type(explanations))
|
81 |
+
print(len(explanations))
|
82 |
|
83 |
print('Done')
|
84 |
|
inference_beit.py
CHANGED
@@ -0,0 +1,203 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import tensorflow as tf
|
2 |
+
gpu_devices = tf.config.experimental.list_physical_devices('GPU')
|
3 |
+
if gpu_devices:
|
4 |
+
tf.config.experimental.set_memory_growth(gpu_devices[0], True)
|
5 |
+
else:
|
6 |
+
print(f"TensorFlow device: {gpu_devices}")
|
7 |
+
|
8 |
+
import os
|
9 |
+
import numpy as np
|
10 |
+
import keras
|
11 |
+
from PIL import Image
|
12 |
+
import keras_cv
|
13 |
+
from keras_cv_attention_models import beit
|
14 |
+
import matplotlib.pyplot as plt
|
15 |
+
|
16 |
+
|
17 |
+
#preprocessing
|
18 |
+
#TODO
|
19 |
+
num_classes = len(class_names)
|
20 |
+
AUTO = tf.data.AUTOTUNE
|
21 |
+
rand_augment = keras_cv.layers.RandAugment(value_range = (-1, 1), augmentations_per_image = 3, magnitude=0.5)
|
22 |
+
|
23 |
+
SIZE = 384
|
24 |
+
debug = None
|
25 |
+
|
26 |
+
def augmentations(x, crop_size=22, brightness=0.2, contrast=0.2, saturation=0.2, hue=0.2):
|
27 |
+
x = tf.cast(x, tf.float32)
|
28 |
+
x = tf.image.random_crop(x, (tf.shape(x)[0], 100, 100, 3))
|
29 |
+
x = tf.image.random_brightness(x, max_delta=brightness)
|
30 |
+
x = tf.image.random_contrast(x, lower=1.0-contrast, upper=1+contrast)
|
31 |
+
x = tf.image.random_saturation(x, lower=1.0-saturation, upper=1.0+saturation)
|
32 |
+
x = tf.image.random_hue(x, max_delta=hue)
|
33 |
+
x = tf.image.resize(x, (128, 128))
|
34 |
+
x = tf.clip_by_value(x, 0.0, 255.0)
|
35 |
+
x = tf.keras.applications.resnet_v2.preprocess_input(x)
|
36 |
+
return x
|
37 |
+
|
38 |
+
|
39 |
+
def pad_gt(x):
|
40 |
+
h, w = x.shape[-2:]
|
41 |
+
padh = sam.image_encoder.img_size - h
|
42 |
+
padw = sam.image_encoder.img_size - w
|
43 |
+
x = F.pad(x, (0, padw, 0, padh))
|
44 |
+
return x
|
45 |
+
|
46 |
+
def preprocess(img):
|
47 |
+
|
48 |
+
img = np.array(img).astype(np.uint8)
|
49 |
+
|
50 |
+
#assert img.max() > 127.0
|
51 |
+
|
52 |
+
img_preprocess = predictor.transform.apply_image(img)
|
53 |
+
intermediate_shape = img_preprocess.shape
|
54 |
+
|
55 |
+
img_preprocess = torch.as_tensor(img_preprocess).cuda()
|
56 |
+
img_preprocess = img_preprocess.permute(2, 0, 1).contiguous()[None, :, :, :]
|
57 |
+
|
58 |
+
img_preprocess = sam.preprocess(img_preprocess)
|
59 |
+
if len(intermediate_shape) == 3:
|
60 |
+
intermediate_shape = intermediate_shape[:2]
|
61 |
+
elif len(intermediate_shape) == 4:
|
62 |
+
intermediate_shape = intermediate_shape[1:3]
|
63 |
+
|
64 |
+
return img_preprocess, intermediate_shape
|
65 |
+
|
66 |
+
|
67 |
+
|
68 |
+
def normalize(img):
|
69 |
+
img = img - tf.math.reduce_min(img)
|
70 |
+
img = img / tf.math.reduce_max(img)
|
71 |
+
img = img * 2.0 - 1.0
|
72 |
+
return img
|
73 |
+
|
74 |
+
def smooth_mask(mask, ds=20):
|
75 |
+
shape = tf.shape(mask)
|
76 |
+
w, h = shape[0], shape[1]
|
77 |
+
return tf.image.resize(tf.image.resize(mask, (ds, ds), method="bicubic"), (w, h), method="bicubic")
|
78 |
+
|
79 |
+
def resize(img):
|
80 |
+
# default resize function for all pi outputs
|
81 |
+
return tf.image.resize(img, (SIZE, SIZE), method="bicubic")
|
82 |
+
|
83 |
+
def pi(img, mask):
|
84 |
+
img = tf.cast(img, tf.float32)
|
85 |
+
|
86 |
+
shape = tf.shape(img)
|
87 |
+
w, h = tf.cast(shape[0], tf.int64), tf.cast(shape[1], tf.int64)
|
88 |
+
|
89 |
+
mask = smooth_mask(mask)
|
90 |
+
mask = tf.reduce_mean(mask, -1)
|
91 |
+
|
92 |
+
img = img * tf.cast(mask > 0.1, tf.float32)[:, :, None]
|
93 |
+
|
94 |
+
img_resize = tf.image.resize(img, (SIZE, SIZE), method="bicubic", antialias=True)
|
95 |
+
img_pad = tf.image.resize_with_pad(img, SIZE, SIZE, method="bicubic", antialias=True)
|
96 |
+
|
97 |
+
# building 2 anchors
|
98 |
+
anchors = tf.where(mask > 0.15)
|
99 |
+
anchor_xmin = tf.math.reduce_min(anchors[:, 0])
|
100 |
+
anchor_xmax = tf.math.reduce_max(anchors[:, 0])
|
101 |
+
anchor_ymin = tf.math.reduce_min(anchors[:, 1])
|
102 |
+
anchor_ymax = tf.math.reduce_max(anchors[:, 1])
|
103 |
+
|
104 |
+
if anchor_xmax - anchor_xmin > 50 and anchor_ymax - anchor_ymin > 50:
|
105 |
+
|
106 |
+
img_anchor_1 = resize(img[anchor_xmin:anchor_xmax, anchor_ymin:anchor_ymax])
|
107 |
+
|
108 |
+
delta_x = (anchor_xmax - anchor_xmin) // 4
|
109 |
+
delta_y = (anchor_ymax - anchor_ymin) // 4
|
110 |
+
img_anchor_2 = img[anchor_xmin+delta_x:anchor_xmax-delta_x,
|
111 |
+
anchor_ymin+delta_y:anchor_ymax-delta_y]
|
112 |
+
img_anchor_2 = resize(img_anchor_2)
|
113 |
+
else:
|
114 |
+
img_anchor_1 = img_resize
|
115 |
+
img_anchor_2 = img_pad
|
116 |
+
|
117 |
+
# building the anchors max
|
118 |
+
anchor_max = tf.where(mask == tf.math.reduce_max(mask))[0]
|
119 |
+
anchor_max_x, anchor_max_y = anchor_max[0], anchor_max[1]
|
120 |
+
|
121 |
+
img_max_zoom1 = img[tf.math.maximum(anchor_max_x-SIZE, 0): tf.math.minimum(anchor_max_x+SIZE, w),
|
122 |
+
tf.math.maximum(anchor_max_y-SIZE, 0): tf.math.minimum(anchor_max_y+SIZE, h)]
|
123 |
+
|
124 |
+
img_max_zoom1 = resize(img_max_zoom1)
|
125 |
+
img_max_zoom2 = img[anchor_max_x-SIZE//2:anchor_max_x+SIZE//2,
|
126 |
+
anchor_max_y-SIZE//2:anchor_max_y+SIZE//2]
|
127 |
+
img_max_zoom2 = img[tf.math.maximum(anchor_max_x-SIZE//2, 0): tf.math.minimum(anchor_max_x+SIZE//2, w),
|
128 |
+
tf.math.maximum(anchor_max_y-SIZE//2, 0): tf.math.minimum(anchor_max_y+SIZE//2, h)]
|
129 |
+
#tf.print(img_max_zoom2.shape)
|
130 |
+
#img_max_zoom2 = resize(img_max_zoom2)
|
131 |
+
|
132 |
+
return tf.cast(img_resize, tf.float32)
|
133 |
+
|
134 |
+
def parse_img(element, split, randaugment,maskaugment=True):
|
135 |
+
#global debug
|
136 |
+
path, class_id = element[0], element[1]
|
137 |
+
|
138 |
+
data = tf.io.read_file(path)
|
139 |
+
img = tf.io.decode_jpeg(data)
|
140 |
+
img = tf.cast(img, tf.uint8)
|
141 |
+
img = normalize(img)
|
142 |
+
shape = tf.shape(img)
|
143 |
+
|
144 |
+
# data_mask = tf.io.read_file(path_mask)
|
145 |
+
# mask = tf.io.decode_jpeg(data_mask)
|
146 |
+
|
147 |
+
class_id = tf.strings.to_number(class_id)
|
148 |
+
class_id = tf.cast(class_id, tf.int32)
|
149 |
+
|
150 |
+
label = tf.one_hot(class_id, num_classes)
|
151 |
+
|
152 |
+
# img = pi(img, mask)
|
153 |
+
img = tf.image.resize_with_pad(img, SIZE, SIZE, method="bicubic", antialias=True)
|
154 |
+
|
155 |
+
return tf.cast(img, tf.float32), tf.cast(label, tf.int32)
|
156 |
+
|
157 |
+
SIZE = 384
|
158 |
+
wsize=hsize=SIZE
|
159 |
+
def resize_images(batch_x, width=224, height=224):
|
160 |
+
return tf.image.resize(batch_x, (width, height))
|
161 |
+
|
162 |
+
def load_img(image_path,gray=False):
|
163 |
+
img = tf.io.read_file(image_path)
|
164 |
+
img = tf.image.decode_jpeg(img, channels=3)
|
165 |
+
img = tf.image.convert_image_dtype(img, tf.float32)
|
166 |
+
if gray:
|
167 |
+
img = tf.image.rgb_to_grayscale(img)
|
168 |
+
img = tf.image.grayscale_to_rgb(img)
|
169 |
+
img = tf.image.resize(img,(wsize,hsize))
|
170 |
+
return img
|
171 |
+
|
172 |
+
LR = 1e-3
|
173 |
+
|
174 |
+
optimizer = tf.keras.optimizers.Adam(LR)
|
175 |
+
cce = tf.keras.losses.categorical_crossentropy
|
176 |
+
|
177 |
+
model_path = '/content/drive/MyDrive/Gg_Fossils_data_shared_copy/Fossils/models/model-13.h5'
|
178 |
+
model = keras.models.load_model(model_path, custom_objects = {'cce': cce})
|
179 |
+
|
180 |
+
outputs = model.predict(images)
|
181 |
+
|
182 |
+
predictions = tf.math.top_k(outputs[1], k = 5)
|
183 |
+
cid = 1
|
184 |
+
dataset = np.array(dataset)
|
185 |
+
final_predictions = []
|
186 |
+
for ele in predictions[1]:
|
187 |
+
if cid in ele:
|
188 |
+
final_predictions.append(cid)
|
189 |
+
else:
|
190 |
+
final_predictions.append(cid+10)
|
191 |
+
final_predictions = np.array(final_predictions)
|
192 |
+
images2 = images[final_predictions == cid]
|
193 |
+
image2_paths = dataset[final_predictions == cid][:,0]
|
194 |
+
print(images2.shape)
|
195 |
+
|
196 |
+
def get_beit_model(input_shape, num_labels, load_weights=False, ...):
|
197 |
+
pass
|
198 |
+
|
199 |
+
def inference_dino(input_image, model_name):
|
200 |
+
pass
|
201 |
+
|
202 |
+
def inference_beit_embedding(input_image, model, size=600):
|
203 |
+
pass
|