raphael/siamese
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

adds preloading weights into existing siamese network

Browse Source
This commit is contained in:
Raphael Maenle 2021-08-12 15:25:37 +02:00
parent d6e6abc8af
commit d8d2b5463e
2 changed files with 21 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
evaluate.py
View File

@ -1,21 +1,25 @@
import tensorflow.keras as keras import tensorflow.keras as keras
from PIL import Image from PIL import Image
import numpy as np import numpy as np
import pdb import ipdb
def getI(path): class g:
return np.asarray(Image.open(path).convert('RGB').resize((100, 100))) / 255 def __init__(self, path):
self.image = np.asarray(Image.open(path).convert('RGB').resize((100, 100))) / 255
def show(self):
self.image.show()
def predict(image1, image2): def predict(image1, image2):
return model.predict([np.array([image2]), np.array([image1])]) return model.predict([np.array([image2.image]), np.array([image1.image])])
model = keras.models.load_model('./siamese_checkpoint') model = keras.models.load_model('../siamese_100x100_pretrainedb_vgg16')
image1 = getI('data/fruits/fruits-360/Training/Avocado/r_254_100.jpg') image1 = g('data/fruits/fruits-360/Training/Avocado/r_254_100.jpg')
image2 = getI('data/fruits/fruits-360/Training/Avocado/r_250_100.jpg') image2 = g('data/fruits/fruits-360/Training/Avocado/r_250_100.jpg')
print(predict(image1, image2)) print(predict(image1, image2))
pdb.set_trace() ipdb.set_trace()

11
train_lambda_coco.py
View File

@ -12,8 +12,15 @@ from siamese import SiameseNetwork
import pdb import pdb
import os, math, numpy as np import os, math, numpy as np
# import Image handling and
# set do not import metadata
import PIL
from PIL import Image from PIL import Image
PIL.ImageFile.LOAD_TRUNCATED_IMAGES=True
batch_size = 128 batch_size = 128
num_classes = 131 num_classes = 131
@ -21,9 +28,9 @@ num_classes = 131
img_rows, img_cols = 100, 100 img_rows, img_cols = 100, 100
def createTrainingData(): def createTrainingData():
base_dir = './classified/' base_dir = './data/combined/'
train_test_split = 0.7 train_test_split = 0.7
no_of_files_in_each_class = 200 no_of_files_in_each_class = 200
#Read all the folders in the directory #Read all the folders in the directory
folder_list = os.listdir(base_dir) folder_list = os.listdir(base_dir)