标签： 南京桑拿

The Hornets officially thanked Jordan： Unforgettable 13 years of being spit out and losing 600 games+only 3 times in the playoffs.

On August 5 th, Beijing time, Charlotte Hornets officially took photos of Jordan’s speech as the boss, and thanked Jordan for his 13 years of dedication to the team. "Thank you, Michael Jordan, for an unforgettable 13 years!" Jordan has sold most of his shares to become a minority shareholder of the team, but he has only advanced to the playoffs three times in his 13 years as a big boss. All round trips are quite embarrassing, and he lost 600 regular season games in total.

The Hornets had previously officially announced that Jordan had completed the ownership transaction, and he only kept a small number of shares to become a minority shareholder, while Gaby Plotkin and Rick Schnarr led the group to become the new boss of the Hornets, and Jordan was able to cash in $3 billion.

Now the Hornets officially thanked Jordan. They took photos of Jordan speaking in a suit and wrote "Thank you, Michael Jordan, for an unforgettable 13 years!"

Just for the "unforgettable 13 years", public opinion has different views, especially in the era when Tutu Jordan managed the Hornets. They were recognized as a super fishing team.

In the management era of Jordan as the big boss, the Hornets lost a total of 600 regular-season games, while in the 13 years of Jordan’s career, they lost only 291 regular-season games. The contrast gap is extremely obvious.

What’s more, Jordan’s 13-year career with 6 championships FMVP is recognized as the first person in history, while the Hornets only advanced to the playoffs three times in 13 years, and all of them swam out in one round, which is undoubtedly an unforgettable 13 years.

Stop spending money to rob Kane! Manchester United used to be a bully： they will regret it in two years.

Andy Cole, a famous Manchester United striker, opposed the possibility of introducing Harry Kane to the team and insisted that the British star was a "short-term solution".

Andy Cole believes that Manchester United need a longer-term solution instead of chasing Harry Kane too hard.

The former Manchester United striker said: "Manchester United can no longer solve problems with short-term solutions." They can’t spend 80 million pounds or 100 million pounds on Harry Kane. Then two years later, they will look back and say to themselves, he is 32 years old and we need to find another striker. "

"They can’t put themselves in that position again."

Old strikers United have signed in recent years include Ibrahimovic, 34, in 2016, cavani, 33, in 2020 and Cristiano Ronaldo, 36, in 2021. Andy Cole’s exhortation seems to be talking about Harry Kane’s problems, but from whom did he draw this conclusion?

[A little information] 442 comments on the top 10 centers in today’s football： Harland No.1, Mbappé Benzema Kane No.2.

Live on May 20 th, 442 magazine selected the top ten centers in today’s football. The position on the field is limited to the center, and the player who plays the main wing is not in the selection.

1. Manchester City, Harland

2.Paris, Mbappé

3. Benzema Real Madrid

4. Kane Hotspur

5. Osman Naples

6.levambasa

7. lautaro Inter Milan

8.Arsenal, Jesus

9. Isak Nuka

10. Juventus, Vlahovic

（beltalowda）

The most expensive lineup in football history： Messi without Ronaldo! Mbappé’s 200 million top partner Harland

On April 23rd, Beijing time, the German "Transfer Market" took stock of the most expensive 11-man lineup in history, with a total value of 1.5 billion euros! Let’s see who is selected.

In the goalkeeper position, oblak, the goalkeeper of Atletico Madrid in 2018-19 season, is worth 100 million euros! The left-back is Bayern’s flying wing Alfonso Davis in the 2020-21 season, and the right-back is Liverpool’s main force Arnold in the 2019-20 season, with 110 million yuan. The central defenders are Fan Dike (100 million) in 2019-20 season and Vallanet (80 million) in 2018-19 season.

In midfield, Deborah, the core of Manchester City in the 2017-18 season, is worth 150 million. And this season’s Dortmund star Bellingham, reached 120 million euros. On the left is Neymar, who is worth 180 million euros in the 2017-18 season!

Of course, the right winger is Messi. In the 2017-18 season, the value of the ball king is also 180 million. It is worth mentioning that Cristiano Ronaldo, another peerless and arrogant player, was not selected for this lineup.

On the front line, Harland, the Manchester City center this season, is worth 170 million euros, while Mbappé, the great Paris, is worth 200 million euros in the 2018-19 season, which is also the highest in history.

Leave a message to send points ｜ Copyright of AI works

In the past two years, while AI painting has entered the public’s field of vision, it has also caused a lot of controversy.

In 2022, the AI painting "Space Opera" won the first place in the art competition at the Colorado Expo in the United States, which quickly aroused people’s attention to AI painting, which also triggered a series of discussions such as "Can AI participate in the competition" and "Can AI replace human painters".

AI painting is changing with each passing day and developing rapidly. While it is popular, it has been resisted by many artists because of the problem of "feeding pictures". At the end of last year, Artstation, a world-renowned art website, was screened by an icon boycotting AI works, and artists expressed their anger. At the beginning of this year, another artist filed a class action lawsuit against AI companies such as Stability AI, Midjourney and Deviant Art, claiming that AI companies violated the rights of millions of artists.

Controversy continues because legal progress has not kept pace with the development of science and technology, the current legal norms lack the identification of the ownership of artificial intelligence creations, and the copyright issue of AI paintings cannot be finalized. However, a recent letter from the US Copyright Office has aroused widespread concern.

Screenshot of letters from the US Copyright Office

In February this year, the US Copyright Office re-examined a work that was registered last year. This work is a comic book "Zarya of the Dawn" created by an artist named Kristina Kashtanova using the AI painting tool Midjourney. In a letter to Kristina Kashtanova’s attorney, the US Copyright Office made a new administrative decision: cancel the original copyright registration of Zarija at Dawn and re-register the comic book as a compilation. The reason is that Kristina used the AI tool to create the picture part of the cartoon, but she didn’t explain it when she first applied. The US Copyright Office thinks that this kind of AI painting is not a work created by a human author, and it is not within the scope of "works" protected by law, so it can’t be registered.

The cover and some illustrations of Zarija at Dawn

In the letter, the US Copyright Office emphasized that the protection scope of The Copyright Act is original works, and originality includes independent creativity and full creativity.

Kristina Kashtanova argued that she constantly revised the prompt to achieve the final desired result, and that she also PS-processed the generated pictures. In this regard, the US Copyright Office responded: constantly improving the picture by modifying the prompt is not creation. Kristina Kashtanova’s prompt expression is very short, and the main creative process is completed by AI. What she does is selection rather than creation. In addition, she did too little PS processing on the picture to be considered as creation.

In addition, the Copyright Administration also pointed out that people can’t predict the specific output results of the Midjourney tool. This lack of "creative control" and "dominance" proves that the descriptions and instructions that Kristina Kashtanova only inputs in the AI tool are only "suggestions", and the final generated pictures are still in the final decision of AI, so the works do not contain human creativity and are not "original", which further proves that the Midjourney in this incident is.

If the "uncontrollable" problem emphasized by the US Copyright Office is solved, will the works generated by AI paintings be copyrightable?For example, the ControlNet model of the recent fire has realized the precise control of AI generation to some extent through various image analysis methods such as attitude recognition Openpose, Depth detection and edge detection Canny.

In this regard, we specially consulted Xiao Sa, a senior partner of Beijing Dacheng Law Firm. She explained: "The uncontrollable generation of pictures is one of the reasons and basis for the US Copyright Office to determine that AI works are not original, but precise control is not the essence of the copyright of AI works.The key to whether the product of ControlNet is copyrightable lies in whether it can embody the "originality" of human beings sufficiently and clearly.

If it can be proved that the works generated by AI are original, will they be protected by copyright?

Lawyer Xiao Sa: "It will be protected by copyright. The court of our country has made such a judgment-a Goose Factory Computer System Co., Ltd. v. a company in Shanghai for a dispute over copyright ownership. Tencent developed an AI software called Dreamwriter, which can automatically generate news articles according to the latest situation of the stock market. The defendant reprinted the article without permission, causing disputes. In the judgment, Shenzhen Nanshan Court held that the AI article involved was a work in the sense of Copyright Law: "The creation of the work reflects the originality of a goose factory team in data input, triggering specific conditions, language style selection, etc., and it is copyrightable and should be protected by copyright law."

If the work generated by AI cannot be proved to be original, can the author not enjoy the rights to the work?

If we can’t prove that the AI work is original enough (or the administrative organ doesn’t recognize this originality), we can try to combine the AI work with other protected works. Taking the ruling of the U.S. Copyright Office as an example, aside from the controversy over the copyright of AI works, the Copyright Office has given a realistic solution to the problem that AI paintings are not protected by law-combining with other protected works. For example, comic books involved in the case, although the pictures are not protected by copyright law, the written works (story content) embedded in the pictures belong to works in the sense of copyright law, and both of them can be protected in the form of compiled works, and the author still enjoys the rights to the works.

Some anti-AI groups think that the ruling of the US Copyright Office denies the copyright of AI paintings, but is this really the case?

At the end of the letter, the US Copyright Office specially emphasized that they did not completely rule out the possibility of accepting the copyright registration of AI painting, but only decided that the author’s influence on the image in this case did not reflect enough creativity. In other words, the ruling of the Copyright Administration is only the first time to demarcate the judging boundary of whether AI works are copyrightable at the official level, that is, human originality.

With the change of AI painting technology, it has become a "past tense" for Kristina Kashtanova to rely solely on text information to generate works. Nowadays, many AI painting lovers will build their own models and use the database to train AI, such as various Lora models that are now self-trained on Stable Diffusion. This kind of behavior seems to greatly satisfy the copyright condition of "originality", but it also brings about cliche problems.Does using AI constitute plagiarism? Will it constitute infringement?

Lawyer Xiao Sa gave a detailed explanation: "Whether AI painting infringes is a very complicated issue. In judicial practice, Chinese courts generally follow the standard of "contact+substantial similarity" to judge whether a work is plagiarized, that is, two conditions need to be met: (1) the author of the later work has "contacted" the former work (it is only necessary to prove that it is "likely to contact", for example, the former work has been published by the author on the Internet); (2) The composition of the post-work and the pre-work is substantially similar (abstract observation and overall observation are used for comprehensive judgment on substantive similarity).

We believe that AI works are difficult to be identified as "plagiarized" works under this judgment rule. On the one hand, it is impossible to prove that AI tools have "touched" infringed works by proving that the works have been published in public, as human works do, because AI painting tools are a product based on the combination of massive original data and specific algorithms, and only when the massive original pictures used by AI painting tools to train large models contain "infringed" works can they be "touched"; On the other hand, AI works are very random. Although there may be similarities in painting style and theme with other works, this similarity belongs to the similarity in thought rather than the similarity in expression protected by copyright. Therefore, it is difficult to prove that there is a "substantial similarity" between the two works, whether it is abstracting its single element and comparing them with the infringed works or comparing them from the overall impression.

Lawyer Xiao Sa believes that high probability does not constitute infringement. Generally speaking, AI creation does not infringe the rights of works and authors. As mentioned above, the object protected by the copyright system itself is the expression of original ideas. In other words, it focuses on the expression rather than the thought itself, and the thought, as a subjective thing that has not actually happened, is not within the protection scope of the copyright system. Only this kind of thought exists objectively as an expression and is original can it exist as the object of copyright protection. Therefore, in the field of copyright law, expressions cannot be copied, but ideas can be used for reference. Painting style and subject belong to "thought" rather than "expression". In this sense, the imitation of painting style belongs to the reference of thought rather than plagiarism, so it certainly does not involve the infringement of copyright. For example, birds are also sketched. A picture of a flying bird created by Xiao Zhang is of course different from a picture of a bird perched in a tree created by Xiao Li. Both of them are also within the scope of the works, but there is no infringement relationship between them.

Whenever new technologies and tools appear, they are bound to be questioned. In the future, we may still need to conduct more in-depth legal and ethical research and discussion on the copyright ownership of AI’s creative works. But we have to admit that the trend is irresistible, and it is almost impossible to get the upper hand against a technology with great commercial prospects and practical application that can significantly improve production efficiency.

It is worth noting that in the publicity of the rule of law in Weibo, the official city of the Supreme People’s Procuratorate, China, pictures generated by AI painting tools have been used.

Project Report ｜ Based on ToF technology +AI algorithm, ＂Looking for Blue City＂ launched the ＂Little Blue Lion＂ to create a standard product for intelligent old-age care and fall prevention.

Text | Shen Xiao

Editor | Wang Yutong

China is accelerating into an aging society. According to the data of the Ministry of Civil Affairs, by the end of 2021, the proportion of elderly people aged 60 and over in China was 18.9% of the total population, and it is estimated that by 2025 and 2035, this number will exceed 20% and 30% respectively. With the acceleration of demographic changes, the urgency of providing for the aged is prominent.

However, China’s old-age service market is still facing the gap between supply and demand of caregivers and places of care. In this context, the state and local governments have successively issued a series of policies to guide the development of the old-age care industry, such as 9073 and 9064, encouraging home-based care (90%), supplemented by community day care (7%/6%) and institutional care as a service supplement (3%/4%). However, due to the gradual reduction in the size of domestic family units in recent years, it is difficult to effectively guarantee the manpower needed for home-based care for the elderly. Therefore, it is still necessary to seek other ways to solve the gap between supply and demand of old-age care.

This provides an opportunity for the popularization and application of scientific and technological products in the pension industry. The smart pension industry has emerged, and enterprises have emerged to develop smart devices for different pension application scenarios and provide health monitoring, data analysis and other services based on technologies such as AI and big data.

In the scene where the elderly need care, falling prevention is considered to be a key link. According to the data released by China Disease Surveillance System, falling is the leading cause of injury-related death among the elderly over 65 years old in China.

In order to prevent the elderly from falling, Shanghai Vision Blue City Information Technology Co., Ltd. (hereinafter referred to as Vision Blue City) launched a series of 3D intelligent human posture sensor "Little Blue Lion" based on ToF radar and AI algorithm, and built a corresponding AIoT platform to provide intelligent care services such as posture warning, health data management and analysis.

Through ToF(Time of Flight) sensor, the electronic equipment associated with it can collect the depth data of environment, objects and human body in the scene, and support real-time imaging. With the help of the algorithm, these data can be used to measure the distance, size and track the movement. The AI algorithm expert model and deep learning network based on ToF depth data can realize accurate recognition of human postures such as falling, sitting, lying and walking. At present, ToF technology has been widely used in face recognition and gesture judgment of smart phones. Other application scenarios include: obstacle recognition of sweeping robot; 3D photography and 3D modeling of intelligent hARdware devices such as ar glasses; Unmanned obstacle identification and ranging.

In addition to the "Little Blue Lion" series products, Trendy Blue City has also developed a smart and healthy SaaS platform to provide integrated intelligent care solutions for residential communities, pension real estate, government and private pension institutions, and home-based pension scenarios. The platform covers daily entertainment life management, health management, home care management, etc., and can integrate health data and service early warning information fed back from AIoT platform.

Vision Blue City was established in 2022 with the joint investment of Shanghai Vision Information Technology Co., Ltd. (hereinafter referred to as Vision Technology) and Zhejiang Blue City Chunfeng Construction Management Co., Ltd. (hereinafter referred to as Blue City Chunfeng).

Vision Technology focuses on AI visual behavior analysis technology, and related products and solutions have been applied in retail, security, smart communities and other fields. Blue City Chunfeng belongs to Blue City Group, a comprehensive real estate service provider. Blue City Group has more than 20 years of experience in real estate development and operation management, with blue city agriculture, blue city support and blue city health as its core service industries.

In terms of products, Shan Yiyong, CEO of Trendy Blue City, told 36Kr: "At present, millimeter-wave radar is the main anti-fall technology product used in the smart old-age scene on the market. Compared with RGB cameras, millimeter-wave radar can identify height changes and protect the privacy of the elderly. However, the accuracy of millimeter-wave radar identification is poor, and the probability of attitude false alarm is high. ToF radar can significantly improve the accuracy of attitude reporting while protecting the privacy of the elderly. "

In addition, Vision Blue City will launch the whole industrial chain products based on the combination of body recognition and health care, aiming at systematically collecting human micro-expressions and all kinds of movement data, and using AI technology to realize comprehensive health assessment, rehabilitation and sports guidance, service demand reporting and other functions.

On the one hand, the accuracy and privacy of "Little Blue Lion" series products are double guaranteed because ToF scheme itself has certain technical advantages compared with other technical routes such as millimeter wave radar and laser radar. At the same time, in recent years, the hardware module of ToF sensor has made some progress in technology. On the other hand, it benefits from the technology accumulation of vision recognition AI algorithm.

First of all, the technical advantage lies in the higher privacy protection of imaging, but the collected posture data is clearer and richer. On this basis, supplemented by three-dimensional spatial data,Vision Blue City can use AI technology to make a more accurate judgment and analysis of the user’s posture and behavior information.. Shan Yiyong, CEO of Vision Blue City, said: "The low accuracy of millimeter-wave radar is mainly due to the low amount of data available at present, which can only be used for mechanical judgment and it is difficult to use AI technology. And each of our ToF devices has an AI chip, which can be directly judged. "

Compared with the imaging of other radar sensors

Blue ToF-3D sensor imaging

At present, the ToF sensor applied in Blue City has a maximum distance of 8 meters and a resolution of 640*480. Shan Yiyong, CEO of Vision Blue City, pointed out: "It is very important to improve the collection distance and resolution. In 2021, the ToF sensor module can only support the acquisition distance of 6 meters. By 2022, this number has increased to 8-10 meters. In addition, at the resolution level, it has reached a level similar to VGA, which is enough to clearly distinguish human posture and behavior. "

Secondly, the improvement of the accuracy of body recognition needs the support of corresponding AI algorithm and model, and the model is continuously optimized through a large amount of data. In the early stage of vision technology, a large amount of data has been accumulated for deep learning, and it has a mature model.

Looking forward to Blue City can directly call its ready-made model and quickly write the device algorithm..

At present, the technical team led by Dr. Zhong Zhang, the chief scientist, has solved two ToF technical difficulties.

First, through data modulation, the remote depth data drift can be controlled. According to reports, in the traditional sense, after the monitoring distance reaches 8 meters, the drift distortion of depth data will be more serious;

Secondly, based on deep learning algorithm and expert model, the accuracy of posture recognition is improved. Shan Yiyong, CEO of Vision Blue City, told 36Kr: "There are scenes where deep learning can’t achieve accurate recognition, such as scenes where human feature data is completely or partially lost due to the covering or blocking of obstructions. For example, the elderly may fall behind the coffee table, and ToF equipment can’t capture the portrait behind the coffee table. At this time, the traditional deep learning system will have false positives, but we can infer through the memory and analysis ability of the expert system. "

In the view of blue city, this expert system is also the core technical barrier of the company at present, which can help it build a certain first-Mover advantage.. Shan Yiyong, CEO of Blue City, pointed out: "There are still few people doing things with this technology, and some people do it, but the landing is slow."

On the market side, relying on the resources of Blue City Group and its affiliated enterprises in the real estate industry, it is tending to see that Blue City has signed some old-age real estate projects, such as Hangzhou Blue City Taoranli, etc., with CCRC community and high-end old-age care institutions as the main formats. It is estimated that there will be more than 10 landing projects this year, and mass production will be realized in May, and about 10,000 products will be delivered to customers.

But at the same time, by looking at the technical advantages and resource advantages that Blue City has at this stage, it may face two corresponding challenges at the market level.

First, the product price is high, while the market’s willingness to pay and purchasing power are still low.

From the cost side, Shan Yiyong, CEO of Vision Blue City, said that every ToF device is equipped with AI chips, which leads to the fact that the price of ToF devices cannot be directly reduced to the same cost level as ordinary radar sensors. From the demand side, market education and fee payment are considered as two obstacles for the domestic pension market to move from just-needed pension service to refined management mode of CCRC.

At present, the equipment of Blue City is priced at 2000 yuan. Usually, one piece of equipment needs to be placed in the toilet, bedroom and living room respectively, that is, a single household needs to pay at least 6000 yuan for the equipment. Shan Yiyong, CEO of Vision Blue City, said frankly that this is one of the main reasons why Vision Blue City should first cut into the market from CCRC community and high-end pension institutions. But at the same time, he also said that with the increase of output in the future, the price of external mining modules will decrease; In addition, through the back-end "module+self-developed motherboard" form, the company is expected to achieve a continuous decline in costs, and the price of a device in the future can be reduced to around 1000 yuan.

Second, it remains to be verified whether the existing industry resources can continue to play an advantage in the process of product penetration to B-end customers, G-end customers and C-end customers except affiliated enterprises.

Generally speaking, the business model of Blue City at this stage is mainly divided into two categories: BtoBtoC and BtoGtoC, and the goal of the next stage is to directly face C-end users.

Under BtoBtoC mode, B-end customers such as CCRC community and pension real estate may pay more attention to the actual economic value that the equipment can create for the project, whether the new project adopts ToF scheme or not, and whether the completed project is replaced by ToF scheme. This, in turn, depends to some extent on whether the value of the equipment can be recognized by C-end users. Specifically, the first is the weight of ToF scheme in the decision-making process of users choosing pension institutions; The second is whether users are willing to pay a premium because of the adoption of ToF scheme.

In addition, the body recognition products of Blue City only solve a small part of the needs of the elderly-real-time care and immediate feedback. In the provision of follow-up services, professional institutions and service personnel are needed to fill in. Therefore, whether the adoption of equipment can actually reduce the demand of nursing staff, improve the efficiency of care, improve the reuse rate of personnel and reduce the cost of care may also constitute a part of the core considerations.

Under the BtoGtoC model, the government pays more attention to the creation of social value. Although Vision Blue City proposed a subscription payment model to solve the problem of high equipment cost, as mentioned above,Whether you can achieve synergy with back-end services may be the key.. This is equally important for the C-end users themselves.

In this way, doing a good job in extending the form of products and services and establishing a reliable partnership with upstream and downstream enterprises in the industrial chain may be the future concern of Blue City.

In this regard, Trendy Blue City said that it would explore the form of "property+"and reach a cooperation with the property company. Trendy Blue City is responsible for the provision and basic operation of smart devices, and the property is responsible for the service output after posture warning. In this mode, residents in the community can also pay in the form of equipment and service subscriptions. However, this cooperation model may face some difficulties in actual implementation, such as the delineation of related responsibilities and the ability of the property to provide related services.

At present, the team is mainly composed of R&D personnel, and a marketing team is being established.

Dr. Zhong Zhang, the founder and chief scientist, graduated from Tsinghua University and worked as the chief scientist of OBJECTVIDEO, the world’s first visual behavior analysis company, for nearly 10 years, leading the team to develop an industry-leading expert algorithm model. At present, based on the deep learning model, Dr. Zhong Zhang has built a mature technical path of "expert model system+deep learning system".

Founder and CEO Shan Yiyong has worked in Kangyang Real Estate Head Enterprise for nearly 20 years. He has participated in the planning, construction and operation of many industry benchmarks Kangyang Real Estate projects, and has also participated in the establishment of health management companies and launched O2O community health management solutions based on IoT.

Dr. Mei Jian, founder and CFO, has more than 20 years of experience in the financial industry and 3 years of entrepreneurial experience in the real estate consulting industry. He used to be the chief information officer of state-owned and joint-stock commercial banks and the president of the investment banking department.

Other core members of the team have many years of experience in artificial intelligence, IOT and other fields.