File size: 4,584 Bytes
29d330e | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 | import numpy as np
import torch
import os
from easyfsl.datasets.few_shot_dataset import FewShotDataset
from PIL import Image
import matplotlib.pyplot as plt
class Jammer_Dataset(FewShotDataset):
def __init__(self,data_root,dataset_list,transform =None,mode = "train"):
self.data_root = data_root
self.transform =transform
self.dataset = dataset_list
self.class_names = np.unique([item[1] for item in self.dataset])
#print(f"dataset classes are -------------------------------------------{self.dataset[1][1]}")
print(f"classes are ====================================================={self.class_names}")
self.class_to_label = {v: k for k, v in enumerate(sorted(self.class_names))}
print(f"mapping is ---------------------------------------------------{self.class_to_label}")
self.labels = self.get_labels()
def __len__(self):
return len(self.dataset)
def __getitem__(self,idx):
samples = str(self.dataset[idx][0])
img_array = np.load(os.path.join(samples))
img_array = torch.from_numpy(img_array)
rgb_tensor = torch.stack((img_array, img_array, img_array), dim=0)
img_array = rgb_tensor.permute(0,2,1)
#img_array = img_array[:, int(npy_number):int(npy_number)+243]
# print(x_row_flip_0.shape)
# fig, axs = plt.subplots(1, 2, figsize=(10, 5))
# # Original Image
# axs[0].imshow(img_array)
# axs[0].set_title('Original Image')
# # Flipped Image
# axs[1].imshow(x_row_flip_0.numpy())
# axs[1].set_title(f"Flipped Image_{aug_num}")
# # Save the figures
# plt.savefig(f"original_image_{idx}.png", bbox_inches='tight')
# plt.savefig(f"flipped_image_{idx}.png", bbox_inches='tight')
# Create a 3-channel tensor by repeating the single-channel data
return img_array.float(), self.labels[idx]
def get_labels(self):
return [self.class_to_label[label] for _, label in self.dataset]
def augment_sample(image_path, transforms):
# Load the image
image = Image.open(image_path)
augmented_image = transforms(image)
augmented_image = transforms.ToTensor()(augmented_image)
return augmented_image
def augment_data(data, augmentation_transforms, augmentation_factor):
augmented_data = []
augmented_labels = []
for filename, label in data:
for _ in range(augmentation_factor):
augmented_sample = augment_sample(filename, augmentation_transforms)
augmented_data.append((augmented_sample, label))
augmented_labels.append(label)
return data + augmented_data
def balance_data(n_shot, n_query, data):
print(f"minimum required sample size is {n_shot} + {n_query}")
# Count the number of samples per label in train_data
label_counts_train = {}
for _, label in data:
if label in label_counts_train:
label_counts_train[label] += 1
else:
label_counts_train[label] = 1
# print(label_counts_train)
# print("--------------------started iterating-------------------------------------")
max_repeat = n_shot+n_query
repeats = {}
for key, value in label_counts_train.items():
num_repeats = max(0, max_repeat - value)
if num_repeats > 0:
repeats[key] = num_repeats
#print(f"the reapets are emmkk=====================>>>>>>{repeats}")
repeated_data = []
for label, num_repeats in repeats.items():
label_data = [(filename, label) for filename, data_label in data if data_label == label]
print(f"current procedure running for label {label}")
if len(label_data) > num_repeats:
repeated_data.extend(label_data[:num_repeats+1])
# print(label_data[:num_repeats])
# print("im working")
else:
# print(num_repeats)
#repeated_data.extend([item for item in label_data for _ in range(num_repeats)])
repeated_data.extend(label_data * (num_repeats+1))
# print("else im working")
# print(repeated_data)
# print(repeats)
# print(repeated_data[0])
data.extend(repeated_data)
balanced_label_counts = {}
for _, label in data:
if label in balanced_label_counts:
balanced_label_counts[label] += 1
else:
balanced_label_counts[label] = 1
print("Label counts in the balanced data:")
print(balanced_label_counts)
return data
|