jiangqy/Customized-DataLoader-pytorch

Customized DataLoader for multi label dataset classification-pytorch implementation

---

Customized DataLoader for multi label classification-[pytorch implementation]

1. Details of file fold:

• data/
• data/train_img/*.jpg
• data/train_img.txt
• data/train_label.txt
• data/test_img/*.jpg
• data/test_img.txt
• data/test_label.txt

2. File description:

file	description
data/train_img/	training image fold
data/test_img/	testing image fold
data/train_img.txt	file name list for training image
data/test_img.txt	file name list for testing image
data/train_label.txt	label list for training image
data/test_label.txt	label list for testing image

3. Running example:

requirements:

torch
torchvision

running example:

python multi_label_classifier.py

output:

Training Phase: Epoch: [ 0][ 0/ 3]	Iteration Loss: 0.693
Training Phase: Epoch: [ 1][ 0/ 3]	Iteration Loss: 0.660
Training Phase: Epoch: [ 2][ 0/ 3]	Iteration Loss: 0.619
Training Phase: Epoch: [ 3][ 0/ 3]	Iteration Loss: 0.596
Training Phase: Epoch: [ 4][ 0/ 3]	Iteration Loss: 0.542
Training Phase: Epoch: [ 5][ 0/ 3]	Iteration Loss: 0.509
Training Phase: Epoch: [ 6][ 0/ 3]	Iteration Loss: 0.467
Training Phase: Epoch: [ 7][ 0/ 3]	Iteration Loss: 0.464
Training Phase: Epoch: [ 8][ 0/ 3]	Iteration Loss: 0.439
Training Phase: Epoch: [ 9][ 0/ 3]	Iteration Loss: 0.377
Training Phase: Epoch: [10][ 0/ 3]	Iteration Loss: 0.329
Training Phase: Epoch: [11][ 0/ 3]	Iteration Loss: 0.324

4. Dataset:

We use the following dataset for our example: link.