Why computer vision teaches machines to see - Model Pipeline Impact

Model Pipeline - Why computer vision teaches machines to see

Computer vision helps machines understand images like humans do. It turns pictures into information so machines can recognize objects, faces, or scenes.

Data Flow - 5 Stages

1Input Image

1 image x 64 x 64 pixels x 3 color channels→Load and resize image to fixed size→1 image x 64 x 64 pixels x 3 color channels

A photo of a cat resized to 64x64 pixels with RGB colors

↓

2Preprocessing

1 image x 64 x 64 x 3→Normalize pixel values from 0-255 to 0-1→1 image x 64 x 64 x 3

Pixel value 128 becomes 0.5019608

↓

3Feature Extraction

1 image x 64 x 64 x 3→Apply convolutional filters to detect edges and shapes→1 image x 62 x 62 x 16 feature maps

Edges of cat ears highlighted in feature maps

↓

4Flattening

1 image x 62 x 62 x 16→Convert 3D feature maps into 1D vector→1 vector x 61504 features

All detected features lined up in one long list

↓

5Classification Layer

1 vector x 61504→Fully connected layer to predict class probabilities→1 vector x 10 classes

Output probabilities like [cat: 0.85, dog: 0.05, ...]

Training Trace - Epoch by Epoch

Loss
1.2 |*       
0.9 | *      
0.7 |  *     
0.5 |   *    
0.4 |    *   
    +---------
     1 2 3 4 5 Epochs

Epoch	Loss ↓	Accuracy ↑	Observation
1	1.2	0.45	Model starts learning basic patterns
2	0.9	0.60	Accuracy improves as edges and shapes are recognized
3	0.7	0.72	Model learns more complex features
4	0.5	0.82	Good feature extraction leads to better predictions
5	0.4	0.88	Model converges with high accuracy

Prediction Trace - 4 Layers

Layer 1: Input Image

Layer 2: Convolutional Layer

Layer 3: Flattening

Layer 4: Fully Connected Layer

Model Quiz - 3 Questions

What does the convolutional layer mainly detect in an image?

A Edges and simple shapes

B Text labels

C Sound patterns

D Raw pixel colors only

Key Insight

Computer vision models learn to see by breaking images into simple patterns like edges, then combining these patterns to recognize objects. Training improves the model's ability to predict correctly by reducing errors and increasing accuracy.