siamese/evaluate.py

import tensorflow.keras as keras
from PIL import Image
import numpy as np
import pdb

def getI(path):
    return np.asarray(Image.open(path).convert('RGB').resize((100, 100))) / 255

def predict(image1, image2):
    return model.predict([np.array([image2]), np.array([image1])]) 

model = keras.models.load_model('./siamese_checkpoint')
image1 = getI('../towards/data/fruits-360/Training/Avocado/r_254_100.jpg')
image2 = getI('../towards/data/fruits-360/Training/Avocado/r_250_100.jpg')

print(predict(image1, image2))

pdb.set_trace()
remodels siamese network with vgg16 and 100x100 input - worse performance than with initial design - vgg16 pretrained weights are used for the base network, which is then piped into a custom head model, which - flattens the layer (previously done in the base model) + Dense Layer + Normalization + Activation - training split with 360 fruits used, same as previous mode - maximum prediction level around 0.95 after ca 60 epochs 2021-07-28 19:02:48 +02:00			`import tensorflow.keras as keras`
			`from PIL import Image`
			`import numpy as np`
			`import pdb`
adds coco image split script, update evaluation - evaluation now pauses via pdb and correctly divides the image sizes - coco image requires a 'coco' folder with the COCO 'val2014' and 'annotations' folder downloaded from the cocodataset website - the script splits up the dataset by snipping the bounding- box from the images and saving it into a seperate folder for later parsing by the siamese network 2021-08-06 13:24:08 +02:00
			`def getI(path):`
			`return np.asarray(Image.open(path).convert('RGB').resize((100, 100))) / 255`

			`def predict(image1, image2):`
			`return model.predict([np.array([image2]), np.array([image1])])`

remodels siamese network with vgg16 and 100x100 input - worse performance than with initial design - vgg16 pretrained weights are used for the base network, which is then piped into a custom head model, which - flattens the layer (previously done in the base model) + Dense Layer + Normalization + Activation - training split with 360 fruits used, same as previous mode - maximum prediction level around 0.95 after ca 60 epochs 2021-07-28 19:02:48 +02:00			`model = keras.models.load_model('./siamese_checkpoint')`
adds coco image split script, update evaluation - evaluation now pauses via pdb and correctly divides the image sizes - coco image requires a 'coco' folder with the COCO 'val2014' and 'annotations' folder downloaded from the cocodataset website - the script splits up the dataset by snipping the bounding- box from the images and saving it into a seperate folder for later parsing by the siamese network 2021-08-06 13:24:08 +02:00			`image1 = getI('../towards/data/fruits-360/Training/Avocado/r_254_100.jpg')`
			`image2 = getI('../towards/data/fruits-360/Training/Avocado/r_250_100.jpg')`

			`print(predict(image1, image2))`
remodels siamese network with vgg16 and 100x100 input - worse performance than with initial design - vgg16 pretrained weights are used for the base network, which is then piped into a custom head model, which - flattens the layer (previously done in the base model) + Dense Layer + Normalization + Activation - training split with 360 fruits used, same as previous mode - maximum prediction level around 0.95 after ca 60 epochs 2021-07-28 19:02:48 +02:00
			`pdb.set_trace()`

adds coco image split script, update evaluation - evaluation now pauses via pdb and correctly divides the image sizes - coco image requires a 'coco' folder with the COCO 'val2014' and 'annotations' folder downloaded from the cocodataset website - the script splits up the dataset by snipping the bounding- box from the images and saving it into a seperate folder for later parsing by the siamese network 2021-08-06 13:24:08 +02:00