AR/VR is often compared as twin brothers. It is generally regarded as a new application layer technology or "smart wearable device". Compared with the "algorithm" label of artificial intelligence, it is not deep enough and has connotation. What is the relationship with artificial intelligence? Does AR belong to the artificial intelligence in our current perception?
In March 2018, the Shanghai Municipal Commission of Economy and Information Technology announced the first batch of special projects to be supported by the city's artificial intelligence innovation and development in 2018. "A total of 19 innovative companies were shortlisted, and Liangfengtai, as an AR company, was also shortlisted for this support project." Liangfengtai staff told reporters that this is not the first time AR companies have been classified as artificial intelligence, but this classification method It is also not common. It is understood that this special project is jointly carried out by the Economic and Information Commission and the Municipal Finance Bureau, and the proposed amount of support exceeds 100 million.
Simply sort out the core technology of ARAR (Augmented Reality) is to superimpose virtual information in the real world, that is, to "enhance" reality. This enhancement can come from sight, hearing and even touch. The main purpose is to make the real world and virtual The world merged together.
Among them, the cognition of the real world is mainly reflected in vision, which requires the use of cameras to help obtain information, and feedback in the form of images and videos. Through video analysis, realize the perception and understanding of the three-dimensional world environment, such as the 3D structure of the scene, what objects are inside, and where in the space. The purpose of 3D interactive understanding is to inform the system what to "enhance".
Figure. Typical AR process
There are several key points:The first is 3D environment understanding. To understand what you see, you mainly rely on object/scene recognition and positioning technology. Recognition is mainly used to trigger AR response, while positioning is to know where to superimpose AR content. Positioning can also be divided into coarse positioning and fine positioning according to different accuracy. Coarse positioning is to give a rough position, such as area and trend. The fine positioning may need to be accurate to the point, such as the XYZ coordinates in the 3D coordinate system, and the angle of the object. According to different application environments, positioning in both dimensions has application requirements in AR. In the AR field, common detection and recognition tasks are human face detection, pedestrian detection, vehicle detection, gesture recognition, biometric recognition, emotion recognition, natural scene recognition, etc.
After perceiving the real 3D world and fusing it with virtual content, it is necessary to present this virtual and real fusion information in a certain way. What is needed here is the second key technology in AR: display technology. At present, most AR systems use perspective Head-mounted display, which is divided into video perspective and optical perspective. Other representatives include light field technology (mainly famous for Magic Leap), holographic projection (which often appears in science fiction film and television drama works), etc.
The third key technology in AR is human-computer interaction, which allows people to interact with the superimposed virtual information. AR pursues natural human-computer interaction methods other than touch buttons, such as voice, gestures, gestures, and faces. Use more voice and gestures.
The technical connection between artificial intelligence and ARThere are several concepts often mentioned in the field of artificial intelligence, such as deep learning (DL) and machine learning (ML). In the academic field, several fields including artificial intelligence (AI) have their own research boundaries. In the sense, we often talk about artificial intelligence in general, encompassing all the application directions of technology that "make machines look like humans".
From this picture, you can also get a simple glimpse of the relationship between the three. Deep learning is a technical way to realize machine learning, and machine learning is to make machines smart and achieve artificial intelligence. It can be said that artificial intelligence is the ultimate goal, and machine learning is a technical direction extended to achieve this goal. Among them, there is another important concept of computer vision (CV), which is mainly to study how to make machines "see" like humans. It is an important branch of the current artificial intelligence concept. This is also because the most important thing for humans to obtain information One of the methods is vision. At present, computer vision has already exerted its value in the commercial market, such as face recognition; reading traffic signals and paying attention to pedestrians in automatic driving for navigation; industrial robots used to detect problems and control processes; three-dimensional environment reconstruction image processing and many more. These concepts have both distinction and overlap.
Among them, beginning in 2006, the deep learning craze initiated by Hinton began to spread, which to a certain extent led to the rise of AI again. In the past ten years, significant achievements have been made in many fields including speech recognition, computer vision, and natural language processing. Breakthroughs and extensions to the application fields are in full swing.
Among the core technologies of AR, 3D environment understanding, 3D interactive understanding, computer vision, and deep learning are closely related. 3D environment understanding in academia mainly corresponds to the field of computer vision, and in recent years deep learning has been widely used in computer vision. In terms of interaction, the use of more natural interaction methods such as gestures and voice in hardware terminals has benefited from the breakthroughs in deep learning in related fields in recent years. It can also be said that the application of deep learning in AR is mainly in key vision technologies.
At present, the most common form of AR is 2D image scanning and recognition, as seen in most AR marketing such as Tencent QQ-AR torch campaign and Alipay Wufu. Scanning recognition images with mobile phones shows superimposed content, but the main research and development direction is still 3D objects Recognition and 3D scene modeling.
Real objects exist in 3D form, with different angles and spatial orientations. So a natural extension is from 2D image recognition to 3D object recognition, to recognize the category and posture of the object, deep learning can be used here. Taking fruit recognition as an example, different types of fruits are recognized, and the positioning area is given, which integrates the functions of object recognition and detection.
3D scene modeling expands from recognizing 3D objects to larger and more complex 3D areas. For example, identifying what is in the scene, their spatial location and mutual relationship, etc. This is 3D scene modeling, which is the core technology of AR. This involves the currently popular SLAM (real-time positioning and map construction). By scanning a certain scene, and then superimposing three-dimensional virtual content such as a virtual battlefield on it. If it is only based on ordinary 2D image recognition, a specific picture is required, and the recognition will fail when the picture is not visible. In SLAM technology, even if a specific plane does not exist, the spatial positioning is still very accurate because of the help of the surrounding 3D environment.
Here I want to discuss the integration of deep learning and SLAM technology. Computer vision can be roughly divided into two schools, a learning-based idea, such as feature extraction-feature analysis-classification. At present, deep learning technology has taken the lead in this route. Sexual status. The other kind of route is based on geometric vision, which infers the spatial structure information of objects from lines, edges, and 3D shapes. The representative technology is SFM/SLAM. In the direction of learning, deep learning basically dominates the world, but in the field of geometric vision, there is still very little progress. From academia, the research progress of deep learning technology can be said to be changing with each passing day, while the progress of SLAM technology in the latest decade is relatively small. At the SLAM technology symposium organized by ICCV 2015, the top international vision conference, based on the rapid development of deep learning in other fields of vision in recent years, some experts have proposed the possibility of using deep learning in SLAM, but there is no mature idea yet. . In general, the fusion of deep learning and SLAM in the short term is a direction worth studying. In the long term, combining semantic and geometric information is a very valuable trend. Therefore, SLAM+DL is worth looking forward to.
In terms of interaction methods, the main ones include speech recognition and gesture recognition. Speech recognition has made great progress. Domestic companies such as Baidu, iFLYTEK, and Yunzhisheng are among the best, and AR companies want to make breakthroughs. It is the mature commercialization of gesture recognition.
"A deep learning-based gesture recognition system demonstrated by Liangfengtai mainly defines six gestures up, down, left and right, clockwise, and counterclockwise." Liangfengtai staff told reporters that the detection and positioning of human hands should be realized first, and then passed Recognize the corresponding gesture trajectory to realize the recognition of human gestures. Although other popular areas of artificial intelligence such as face recognition are also used in AR, they are not an important research and development direction of AR companies.
It is not difficult to see from the above that the underlying technology or basic part of AR is the integration of computer vision and related fields, and the combination of the popular deep learning and AR is also the direction of the algorithm engineers. This is also the intersection of computer vision and human-computer interaction. The foundation of AR is the basis of artificial intelligence and computer vision.
Figure: Computer Vision and AR Process Association
In the "Artificial Intelligence Impact Report" released by Toutiao last year, it also simply counted the distribution of artificial intelligence scientists, including companies and large R&D institutions in the fields of face recognition, speech recognition, robotics, AR, and chips. The distribution of R&D personnel also illustrates the direction of segmentation in the AI ​​field.
For AR practitioners, the ideal state is to replace smart phones with smarter AR terminals. Therefore, for users, the first thing that will be affected by the use of AR is the content, followed by the terminal. The AR industry chain can be roughly divided into technology provision. Companies, smart terminal R&D companies, and AR content providers. Among them, AR equipment providers inevitably pay attention to hardware technologies, such as underlying chips, batteries, optical lenses, etc., as well as the performance optimization of the hardware itself, while content providers are more inclined to optimize content and performance based on existing technologies. So we can say that AR technology providers, or AR companies with certain achievements in the research and development of underlying algorithms, are artificial intelligence companies.
For companies, especially startups, they will transform the underlying technology into mature products or services, which may be drones, AR smart terminals, robots, etc., or industry solutions to achieve commercial goals. It has become the expectations and requirements of the media, enterprises and the public for AI companies after the boiling voice. In the near future, the book "Artificial Intelligence Wave: 100 Cutting-Edge AI Applications That Technology Change Life" published by the Artificial Intelligence Industry Development Alliance (AIIA) will be released, and it will cover the cutting-edge achievements of current giants and startups in commercialization. It also directly reflects the current main commercialization direction of AI.
As a technology-driven business field, whether it is AR or artificial intelligence in most other directions, the technology still has a long way to go before it is fully mature. As the entire industry chain is gradually prospering, while focusing on commercialization, more needs to be done. Companies continue to expand the technological boundaries, establish core competitiveness, and allow the industry to explode with greater value and potential. In this way, China's corner overtaking in the AI ​​era is expected.
24V Battery Pack ,Large Battery Pack,24 V Battery Pack,24V Lithium Ion Battery Pack
Zhejiang Casnovo Materials Co., Ltd. , https://www.casnovo-new-energy.com