Task $1.$ This is pytorch task: Find some data set with signatures and train this code to work as it should. Try using the following: Create a custom dataloader in pytorch for your dataset $-.$ given code

Train resnet$50($pytorch documentation$)$ on gallery and try it$,$ evaluate it on tests $-$ for now only gallery and tests.

import zipfile

with zipfile.ZipFile$(\text{'}$some data set from kaggle', $\text{'}$r$\text{'})$ as zip$\_$ref:

$$ zip$\_$ref.extractall$(\text{'}/$content$/$some data set from kaggle'$)$

import pandas as pd

import os $$# Ensure you import os to use it in file paths

from torchvision.io import read$\_$image

from torch.utils.data import Dataset $$# Import Dataset class

$$

class CustomImageDataset$($Dataset$)$:

$$ # the FashionMNIST images are stored in a directory img$\_$dir

$$ # The $\_\_$init$\_\_$ function is run once when instantiating the Dataset object.

$$ # We initialize the directory containing the images, the annotations file, and both transforms

$$ def $\_\_$init$\_\_($self$,$ annotations$\_$file, img$\_$dir, transform$=$None, target$\_$transform$=$None$)$:

$$

$$ #labels are stored separately in a CSV file annotations$\_$file

$$ self.img$\_$labels $=$ pd$.$read$\_$csv$($annotations$\_$file$)$

$$ self.img$\_$dir $=$ img$\_$dir

$$ self.transform $=$ transform

$$ self.target$\_$transform $=$ target$\_$transform

$$

$$ # The $\_\_$len$\_\_$ function returns the number of samples in our dataset.

$$ def $\_\_$len$\_\_($self$)$:

$$ return len$($self$.$img$\_$labels$)$

$$

$$ # The $\_\_$getitem$\_\_$ function loads and returns a sample from the dataset at the given index idx.

$$ # Based on the index, it identifies the image$$s location on disk, converts that to a tensor using read$\_$image,

$$ # retrieves the corresponding label from the csv data in self.img$\_$labels,

$$ # calls the transform functions on them $($if applicable$),$ and returns the tensor image and corresponding label in a tuple.

$$ def $\_\_$getitem$\_\_($self$,$ idx$)$:

$$ img$\_$path $=$ os$.$path.join$($self$.$img$\_$dir, self.img$\_$labels.iloc$[$idx$,0])$

$$ image $=$ read$\_$image$($img$\_$path$)$

$$ label $=$ self.img$\_$labels.iloc$[$idx$,1]$

$$ if self.transform:

$$ image $=$ self.transform$($image$)$

$$ if self.target$\_$transform:

$$ label $=$ self.target$\_$transform$($label$)$

$$ return image, label

$$

from torch.utils.data import DataLoader

$$

train$\_$dataloader $=$ DataLoader$($training$\_$data, batch$\_$size$=64,$ shuffle$=$True$)$

test$\_$dataloader $=$ DataLoader$($test$\_$data, batch$\_$size$=64,$ shuffle$=$True$)$

$$

# Display image and label.

train$\_$features, train$\_$labels $=$ next$($iter$($train$\_$dataloader$))$

print$($f$"$Feature batch shape: $\{$train$\_$features.size$()\}")$

print$($f$"$Labels batch shape: $\{$train$\_$labels.size$()\}")$

img $=$ train$\_$features$[0].$squeeze$()$

label $=$ train$\_$labels$[0]$

plt$.$imshow$($img$,$ cmap$=$"gray"$)$

plt$.$show$()$

print$($f$"$Label: $\{$label$\}")$