Inception modules in Computer Vision - Model Metrics & Evaluation

Metrics & Evaluation - Inception modules

Which metric matters for Inception modules and WHY

Inception modules are used in image recognition tasks. The key metric to check is accuracy, which tells us how many images the model labels correctly. Because Inception modules help the model learn features at different scales, accuracy shows if this helps the model see details better.

Besides accuracy, top-5 accuracy is also important. It checks if the correct label is among the model's top 5 guesses, useful when many classes look similar.

Confusion matrix example

      Actual \ Predicted | Cat | Dog | Bird | Total
      ------------------------------------------
      Cat               | 45  | 3   | 2    | 50
      Dog               | 4   | 40  | 6    | 50
      Bird              | 1   | 5   | 44   | 50
      ------------------------------------------
      Total             | 50  | 48  | 52   | 150
    

This matrix shows how many images of each animal were correctly or wrongly predicted. For example, 45 cats were correctly predicted as cats (true positives for cat), 3 cats were wrongly predicted as dogs (false negatives for cat).

Precision vs Recall tradeoff with examples

Imagine the model detects cats in photos.

• Precision means: Of all images predicted as cats, how many really are cats? High precision means few wrong cat guesses.
• Recall means: Of all actual cat images, how many did the model find? High recall means the model misses few cats.

If the model has high precision but low recall, it rarely says "cat" unless very sure, but misses many cats. If it has high recall but low precision, it finds most cats but also wrongly calls other animals cats.

For Inception modules, balancing precision and recall is important to recognize many objects correctly without too many mistakes.

What good vs bad metric values look like for Inception modules

• Good: Accuracy above 80% on a diverse image set, precision and recall both above 75%, and top-5 accuracy above 90%. This means the model is reliable and finds most objects correctly.
• Bad: Accuracy below 50%, precision or recall below 50%, or top-5 accuracy near random chance (e.g., 20% for 5 classes). This means the model struggles to learn useful features.

Common pitfalls in metrics for Inception modules

• Accuracy paradox: If the dataset is mostly one class, high accuracy can be misleading. The model might just guess the common class.
• Data leakage: If test images are too similar to training images, metrics look better but model won't generalize.
• Overfitting: Very high training accuracy but low test accuracy means the model memorizes training images but fails on new ones.
• Ignoring top-5 accuracy: For many classes, top-1 accuracy alone may not show model usefulness.

Self-check question

Your Inception model has 98% accuracy but only 12% recall on a rare animal class. Is it good for production? Why or why not?

Answer: No, it is not good. The model misses most rare animal images (low recall), even if overall accuracy is high. This means it fails to find important rare cases, which can be critical depending on the application.

Key Result

Accuracy and balanced precision-recall are key to evaluate Inception modules' effectiveness in image recognition.

Practice

1. What is the main purpose of using 1x1 convolutions in an Inception module?

easy

A. To increase the spatial size of the feature maps

B. To add non-linearity without changing dimensions

C. To replace max pooling layers

D. To reduce the number of channels and keep the model efficient

Solution

1. Step 1: Understand the role of 1x1 convolutions

   1x1 convolutions act as channel-wise feature selectors and reduce the number of channels, lowering computation.

2. Step 2: Connect to Inception module efficiency

   By reducing channels before expensive convolutions, the model stays efficient without losing important information.

3. Final Answer:

   To reduce the number of channels and keep the model efficient -> Option D

4. Quick Check:

   1x1 convolutions reduce channels = B [OK]

Hint: 1x1 convs reduce channels to save computation [OK]

Common Mistakes:

• Thinking 1x1 convs increase spatial size
• Confusing 1x1 convs with pooling layers
• Assuming 1x1 convs only add non-linearity

2. Which of the following is the correct way to combine outputs from different branches in an Inception module?

easy

A. Concatenate the outputs along the channel dimension

B. Use max pooling on all outputs

C. Multiply the outputs element-wise

D. Add the outputs element-wise

Solution

1. Step 1: Identify how Inception combines branch outputs

   Inception modules concatenate outputs from different filter branches along the channel axis to keep all features.

2. Step 2: Understand why concatenation is used

   Concatenation preserves all features from each branch, unlike addition or multiplication which mix them.

3. Final Answer:

   Concatenate the outputs along the channel dimension -> Option A

4. Quick Check:

   Outputs concatenated by channels = D [OK]

Hint: Inception outputs join by channel concat, not add [OK]

Common Mistakes:

• Confusing concatenation with element-wise addition
• Thinking outputs are multiplied
• Assuming pooling merges outputs

3. Given this simplified Inception module code snippet, what is the shape of the output tensor?

import torch
import torch.nn as nn

class SimpleInception(nn.Module):
    def __init__(self):
        super().__init__()
        self.branch1 = nn.Conv2d(192, 64, kernel_size=1)
        self.branch2 = nn.Conv2d(192, 128, kernel_size=3, padding=1)
        self.branch3 = nn.Conv2d(192, 32, kernel_size=5, padding=2)
    def forward(self, x):
        b1 = self.branch1(x)
        b2 = self.branch2(x)
        b3 = self.branch3(x)
        return torch.cat([b1, b2, b3], dim=1)

input_tensor = torch.randn(1, 192, 28, 28)
model = SimpleInception()
output = model(input_tensor)
print(output.shape)

medium

A. (1, 224, 32, 32)

B. (1, 64, 28, 28)

C. (1, 224, 28, 28)

D. (1, 224, 28, 28, 3)

Solution

1. Step 1: Calculate output channels per branch

   Branch1 outputs 64 channels, branch2 outputs 128, branch3 outputs 32. Total channels = 64+128+32 = 224.

2. Step 2: Check spatial dimensions and concatenation

   All convolutions use padding to keep spatial size 28x28. Concatenation along channel dimension keeps height and width same.

3. Final Answer:

   (1, 224, 28, 28) -> Option C

4. Quick Check:

   Channels sum to 224, spatial unchanged = A [OK]

Hint: Sum channels from branches, keep spatial size same [OK]

Common Mistakes:

• Adding spatial dimensions instead of channels
• Ignoring padding effects on size
• Misunderstanding concat dimension

4. Identify the error in this Inception module implementation:

class FaultyInception(nn.Module):
    def __init__(self):
        super().__init__()
        self.branch1 = nn.Conv2d(128, 32, kernel_size=1)
        self.branch2 = nn.Conv2d(128, 64, kernel_size=3, padding=1)
    def forward(self, x):
        b1 = self.branch1(x)
        b2 = self.branch2(x)
        return torch.cat([b1, b2], dim=2)

medium

A. Missing padding in branch2 convolution

B. Concatenation dimension should be 1, not 2

C. Input channels to branch1 are incorrect

D. Using nn.Conv2d instead of nn.Conv1d

Solution

1. Step 1: Check concatenation dimension

   In PyTorch, channel dimension is 1. Concatenating along dim=2 (height) is incorrect for Inception outputs.

2. Step 2: Confirm other parts

   Branch2 padding keeps spatial size consistent; input channels match; Conv2d is correct for images.

3. Final Answer:

   Concatenation dimension should be 1, not 2 -> Option B

4. Quick Check:

   Concat along channels = dim 1 [OK]

Hint: Concat outputs along channel dim (1), not height (2) [OK]

Common Mistakes:

• Concatenating along wrong dimension
• Confusing padding with error
• Misreading input channel sizes

5. You want to design an Inception module that balances feature diversity and computational cost. Which combination best achieves this?

hard

A. Use 1x1 convolutions before 3x3 and 5x5 convolutions, then concatenate outputs

B. Use only 5x5 convolutions without 1x1 convolutions to capture large features

C. Use max pooling only and skip convolutions to reduce cost

D. Stack multiple 3x3 convolutions without any 1x1 convolutions

Solution

1. Step 1: Understand feature diversity and cost tradeoff

   Large filters capture diverse features but are costly. 1x1 convolutions reduce channels before large filters to save cost.

2. Step 2: Evaluate options

   Use 1x1 convolutions before 3x3 and 5x5 convolutions, then concatenate outputs uses 1x1 convs to reduce channels before 3x3 and 5x5, balancing diversity and efficiency. Others either ignore cost or diversity.

3. Final Answer:

   Use 1x1 convolutions before 3x3 and 5x5 convolutions, then concatenate outputs -> Option A

4. Quick Check:

   1x1 convs reduce cost + multi-filter concat = C [OK]

Hint: Use 1x1 convs before big filters for efficiency [OK]

Common Mistakes:

• Ignoring 1x1 convs and increasing cost
• Using only pooling loses feature richness
• Stacking without channel reduction wastes resources