What are Convolution filters in a Convolution Networks
In image processing using convolution networks, a kernel, convolution matrix, or mask is a small matrix. It is used for blurring, sharpening, embossing, edge detection, and more. This is accomplished by doing a convolution between a kernel and an image.
In a convolution operation, the filter is slid over the image, and at each position, the sum of the element-wise product of the filter and the portion of the image under the filter is computed. This sum forms a single pixel in the output image. By sliding the filter across the entire image, a new image is created that has been modified according to the characteristics of the filter.
For example, a smoothing filter might average the pixels in a small neighborhood to blur the image, while an edge detection filter might highlight areas of the image where there is a significant change in intensity, indicating an edge.
Convolution filters are a fundamental tool in image processing and are widely used in computer vision tasks, deep learning models for image recognition, and other applications that involve image analysis.
Filters and definition
In this article, here are some conventions that we are following with respect to convolution networks and convolution filters:
We are specifically referring to 2D convolutions that are usually applied on 2 matrix objects such as images. These concepts also apply for 1D and 3D convolutions, but may not correlate directly.
While applying 2D convolutions like 3X3 convolutions on images, a 3X3 convolution filter, in general will always have a third dimension in size. This filter depends on (and is equal to) the number of channels of the input image. So, we apply a 3X3X1 convolution filter on gray-scale images (the number of channels = 1) whereas, we apply a 3X3X3 convolution filter on a colored image (the number of channels = 3).
We will refer to all the convolutions by their first two dimensions, irrespective of the channels. (We are observing the assumption of zero padding).
How do Convolution Filters Operate
A convolution filter passes over all the pixels of the image in such a manner that, at a given time, we take 'dot product' of the convolution filter and the image pixels to get one final value output. We do this hoping that the weights (or values) in the convolution filter, when multiplied with corresponding image pixels, gives us a value that best represents those image pixels. We can think of each convolution filter as extracting some kind of feature from the image.
Therefore, convolutions are done usually keeping these two things in mind -
Most of the features in an image are usually local. Therefore, it makes sense to take few local pixels at once and apply convolutions.
Most of the features may be found in more than one place in an image. This means that it makes sense to use a single kernel all over the image, hoping to extract that feature in different parts of the image.
Now that we have convolution filter sizes as one of the hyper-parameters to choose from, the choice is can be made between smaller or larger filter size.
Filter Size Considerations
Here are the things to consider while choosing the size:
Smaller Filter Sizes | Larger Filter Sizes |
We only look at very few pixels at a time. Therefore, there is a smaller receptive field per layer. | We look at lot of pixels at a time. Therefore, there is a larger receptive field per layer. |
The features that would be extracted will be highly local and may not have a more general overview of the image. This helps capture smaller, complex features in the image. | The features that would be extracted will be generic, and spread across the image. This helps capture the basic components in the image. |
The amount of information or features extracted will be vast, which can be further useful in later layers. | The amount of information or features extracted are considerably lesser (as the dimension of the next layer reduces greatly) and the amount of features we procure is greater. |
In an extreme scenario, using a 1x1 convolution is like considering that each pixel can give us some useful feature independently. | In an extreme scenario, if we use a convolution filter equal to the size of the image, we will have essentially converted a convolution to a fully connected layer. |
Here, we have better weight sharing, thanks to the smaller convolution size that is applied on the complete image. | Here, we have poor weight sharing, due to the larger convolution size. |
Now that you have a general idea about the extraction using different sizes, we will follow this up with an experiment convolution of 3X3 and 5X5:
Smaller Filter Sizes | Larger Filter Sizes |
If we apply 3x3 kernel twice to get a final value, we actually used (3x3 + 3x3) weights. So, with smaller kernel sizes, we get lower number of weights and more number of layers. | If we apply 5x5 kernel once, we actually used 25 (5x5) weights. So, with larger kernel sizes, we get a higher number of weights but lower number of layers. |
Due to the lower number of weights, this is computationally efficient. | Due to the higher number of weights, this is computationally expensive. |
Due to the larger number of layers, it learns complex, more non-linear features. | Due to the lower number of layers, it learns simpler non linear features. |
With more number of layers, it will have to keep each of those layers in the memory to perform back propagation. This necessitates the need for larger storage memory. | With lower number of layers, it will use less storage memory for back propagation |
Conclusion
Based on the points listed in the above table and from experimentation, smaller kernel filter sizes are a popular choice over larger sizes.
Another question could be the preference for odd number filters or kernels over 2X2 or 4X4.
The explanation for that is that though we may use even sized filters, odd filters are preferable because if we were to consider the final output pixel (of next layer) that was obtained by convolving on the previous layer pixels, all the previous layer pixels would be symmetrically around the output pixel. Without this symmetry, we will have to account for distortions across the layers. This will happen due to the usage of an even sized kernel. Therefore, even sized kernel filters aren’t preferred.
1X1 is eliminated from the list as the features extracted from it would be fine grained and local, with no consideration for the neighboring pixels.
Hence, 3X3 works in most cases, and it often is the popular choice.