Vision

Vision - Whether to learn processes, or to make a living from it. By having two elements of binocular vision and stereoscopic viewing, depth perception becomes possible. Binocular vision is defined as a field of view in which both eyes are directed to the same target, and both eyes operate in cooperation (1). Stereo vision is defined as a vision in which two separate images from two eyes are successfully combined in one image in the brain (1). In the classroom, even though the retina received 2D visual information, I learned that I can gain 3D vision through the process of depth perception.

Cafe is a deep learning framework developed by Berkeley Visual and Learning Center (BVLC) and community contributors. It is widely used by computer vision researchers all over the world. It is mainly built for computer vision, but it can also be used for many other deep learning tasks.

This project taught me the advanced nature of computer vision, and one reason to understand the deep learning of computer vision. The reason for this is that in most cases this parameter manually adjusts the parameters so it is not necessary to process these parameters when learning computer vision in detail.

Computer vision is related to deepening understanding from images. This includes video images, but this is just a series of images. Computer vision is an interdisciplinary field, scientists who study human vision and scientists who study vision of artificial systems provide new discoveries. Computer vision includes sophisticated techniques such as image processing and HOG conversion. However, one of the greatest contributors to computer vision comes from the field of deep learning, the latest top performer of the ImageNet competition. However, to design a deeper deep learning model, you need to design a model that is powerful enough to understand the problem consistently after understanding the image that the system may encounter offline.

In the past few years, Deep Learning dominated computer vision and gained the highest score in many missions and related competitions. ImageNet is the most popular and famous among these computer visual competitions. In the ImageNet contest, researchers need to create a model that best classifies certain images in the dataset. In the past few years, deep learning technology has made rapid progress in this game, surpassing human performance. Today we will explore these developments to gain insight into how these advances can be learned through deep learning, from which we can learn, and where we can go from here to review.