합성곱 이미지필터 - 합성곱의 이해, 이미지필터 만들기

This app is a useful learning tool and experimentation platform for people interested in mathematics, image processing, convolutional neural networks (CNNs), and more. This app uses animation to intuitively explain 2D convolution operations used in computer vision and CNN. Even if you are not a major, you can understand intuitively through visual animation, and at the same time, it provides a fun learning experience.

Users can create their own image filters, apply them to various images, and check the effects in real time.

[Main features of the app]

- Visual animation: Provides a visual animation of the process of 2D convolution operation so you can clearly understand it.

- Convolution calculator: You can set various input matrix and kernel matrix values ​​and calculate 2D convolution operations.

- Image filter: Users can check how the applied filter transforms the image to an image filter implemented based on 2D convolution.

- Multiple filter types provided: Various basic filter presets such as edge detection and blurring are provided, and users can select and customize filters.

[Motivation for app development]

Convolution Flow was developed inspired by the difficulty I had in understanding the concept of convolution while studying computer engineering. 2D convolution operations play an important role in computer vision and CNN, but it was not easy to understand them through text or formulas alone. So, we wanted to create a tool that could easily explain the convolution calculation process with visual animations and experiment with application examples such as image filters.

[Images used within the app]

- The sample images used within the app were legitimately created through OpenAI's DALL-E model to allow users to apply and test convolution-based filters, and the images used do not depict real people.

[Feedback]

- If there are any improvements, errors, or features you would like to see added to the app, please send the following email. We will compile your feedback and reflect it in future updates.

- Email: [email protected]

“Understand convolution as an animation and have a new experience creating your own filter!”