Data augmentation in pipeline in TensorFlow - Model Pipeline Trace

Model Pipeline - Data augmentation in pipeline

This pipeline shows how data augmentation helps a model learn better by creating new, varied images from the original ones during training. It makes the model stronger by showing it many different versions of the same image.

Data Flow - 4 Stages

1Original dataset

1000 rows x 32x32x3 images→Load raw images and labels→1000 rows x 32x32x3 images

Image of a cat with label 'cat'

↓

2Data augmentation

1000 rows x 32x32x3 images→Apply random flips, rotations, and brightness changes→1000 rows x 32x32x3 images (augmented)

Flipped cat image, rotated cat image, brighter cat image

↓

3Batching and shuffling

1000 rows x 32x32x3 images→Shuffle data and group into batches of 32→32 rows x 32x32x3 images per batch

Batch of 32 augmented images with labels

↓

4Model input

32 rows x 32x32x3 images→Feed batch into CNN model→32 rows x 10 class probabilities

Batch predictions for 10 classes

Training Trace - Epoch by Epoch

Loss
1.2 |*       
1.0 | *      
0.8 |  *     
0.6 |   *    
0.4 |    *   
    +---------
     1 2 3 4 5
       Epochs

Epoch	Loss ↓	Accuracy ↑	Observation
1	1.2	0.55	Model starts learning with augmented data
2	0.9	0.68	Loss decreases, accuracy improves
3	0.7	0.75	Model learns better features from augmented images
4	0.55	0.82	Training stabilizes with good accuracy
5	0.45	0.87	Model converges with augmented data

Prediction Trace - 4 Layers

Layer 1: Input image

Layer 2: Data augmentation

Layer 3: CNN layers

Layer 4: Dense layer with softmax

Model Quiz - 3 Questions

Why do we apply data augmentation in the training pipeline?

A To create more varied training images and improve model learning

B To reduce the size of the dataset

C To make the images black and white

D To speed up the training by skipping images

Key Insight

Data augmentation helps the model see many versions of the same image, which improves its ability to generalize and recognize objects better. This leads to lower loss and higher accuracy during training.