# Advanced Computer Vision CNN Architectures

Source: [https://docs.qualcomm.com/doc/80-63442-4/topic/advanced-computer-vision-cnn-architectures.html](https://docs.qualcomm.com/doc/80-63442-4/topic/advanced-computer-vision-cnn-architectures.html)

Visualization in GoogleNet Inception and ResNet

Visualization helps in exploring the layers responsible for extracting a specific
            feature. In the process of building a CNN model, visualizing layers is as important as
            calculating the training error (accuracy) and validation error. It also helps in using
            the pre-trained models for transfer learning, by visualizing and embedding only the
            required filters with specified weights for the current requirement. Visualization can
            be applied to heatmaps of class activation, filters and activation functions.

## GoogleNet Inception

In GoogleNet Inception, multiple size filters are applied at the same level along
                with 1×1 convolution. It is not always possible to use the same filter to detect the
                object from an image. When running object detection and feature extraction on the
                images below, for example, it would be necessary to use a larger filter for the
                image on the left (close-up of brown dog) and a smaller filter on the image on the
                right (dog with tree in background).  
![](data:image/png;base64,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)

(For a diagram of the architecture of GoogleNet Inception, see [https://web.cs.hacettepe.edu.tr/~aykut/classes/spring2016/bil722/slides/w04-GoogLeNet.pdf](https://web.cs.hacettepe.edu.tr/~aykut/classes/spring2016/bil722/slides/w04-GoogLeNet.pdf),
                pp. 31-32.)

GoogleNet Inception helps in avoiding overfitting by increasing the width of the
                model relative to its depth. As the depth of the model increases to capture more
                features, the activation function is responsible for keeping the model from being
                too general due to overfitting the training data.

The images below depict general (simple) and overfitted (complex) models.  
![](data:image/png;base64,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)

In GoogleNet Inception, filters with different sizes are applied to a single input
                because, as mentioned in the dog example above, object detection by feature
                extraction varies. Using filters of multiple sizes at the same level ensures that
                the objects of different sizes in different images are easily detected.

Before applying the filters, the input is convolved with a 1×1 matrix, reducing the
                number of computations made on the image. For example, an input of size 14×14×480
                convolving with a filter of size 5×5×48 results in (14×14×48)×(5×5×480) = 112.9
                million computations. But if the image is convolved with 1×1 before convolving with
                3×3, the number of computations drops to (14×14×16)×(1×1×480) = 1.5 million and
                (14×14×48)×(5×5×16) = 3.8 million. A filter concatenation is used to collect all the
                filters of that inception block.

An inception block can also have an auxiliary branch. The auxiliary branches are
                present in the training-time architecture but not in the test-time architecture.
                They introduce more aggressiveness during classification when they encourage
                discrimination in the lower stages of the classifier. They increase the gradient
                signal that gets propagated back and they provide additional regularization.

## ResNet

If the neural network is too deep, the gradient from the loss function will progress
                towards zero, a problem known as “vanishing gradient.” ResNet is used to solve this
                problem by skipping the connections backward from the later layers to the initial
                layers.

ResNet (Residual Network) contains four residual blocks: Layer1, Layer2, Layer3 and
                Layer4. (For a diagram of the architecture of ResNet, see [https://towardsdatascience.com/understanding-and-visualizing-resnets-442284831be8](https://towardsdatascience.com/understanding-and-visualizing-resnets-442284831be8?gi=d64cbe067213).)

Each block contains a different number of residual units that perform a set of
                convolutions in a unique approach. Each block is augmented by max pooling to reduce
                the spatial dimensions.

As illustrated below, there are two types of residual units: baseline residual units
                and bottleneck residual units.
Figure : Baseline residual unit (L) Bottleneck residual unit (R)

![](data:image/jpeg;base64,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)

The baseline residual unit contains two 3×3 convolutions with batch normalization and
                ReLU activations.

The bottleneck residual unit consists of three stacked operations with a series of
                1×1, 3×3 and 1×1 convolution operations. The two 1×1 operations are designed for
                reducing and restoring dimensions. The 3×3 convolution is used to operate on a
                less-dense feature vector. Also, batch normalization is applied after each
                convolution and before ReLU non-linearity.

**Parent Topic:** [CNN Architectures](https://docs.qualcomm.com/doc/80-63442-4/topic/cnn-architectures.html)

Last Published: Jun 24, 2024

[Previous Topic
Computer Vision CNN Architectures](https://docs.qualcomm.com/bundle/publicresource/80-63442-4/topics/computer-vision-cnn-architectures.md) [Next Topic
Vision-based AI Use Cases](https://docs.qualcomm.com/bundle/publicresource/80-63442-4/topics/vision-based-ai-use-cases.md)