Connecting People and Services Through Visual Recognition: Insights from Baidu's Tech Salon

On May 7th, I participated in the #Xierqi Night Talk# mystery listener recruitment event through the Baidu Technology Salon WeChat group and was lucky enough to be selected as the sole mystery listener.

My name is Zhang Hongchao, a senior Java developer at Sohu Changyou, and I'm very passionate about technology. So when I learned I was chosen as the "sole listener" to attend this high-end, sophisticated closed-door technical meeting #Xierqi Night Talk#, I felt like Lady Luck had descended upon me - or rather, Mr. Luck, since I was about to meet several legendary technical experts face-to-face.

The technical "gurus" at the meeting repeatedly emphasized the need to re-establish connections between people and the world , bringing new ways to play in the mobile internet era.

Want to know what these new ways are? Keep reading!

This session featured Baidu Cloud's Chief Architect Hou Zhenyu, Baidu Mobile Cloud's Chief Evangelist Zhang Hui, and Baidu Deep Learning Research Institute (IDL)'s Executive Vice President Yu Kai.

Let me share what left a deep impression on me. Teacher Hou joined Baidu in 2003, and in two months he'll complete 11 years of work. Hou is very approachable. Through private conversations, I learned that over 10 years ago he joined Baidu and led the development of a series of Baidu's star products, including Baidu Tieba (the leader in domestic community products), Baidu Knows, Baidu Space, and Baidu Passport. The 1.0 versions of these products all came from Teacher Zhenyu's hands. Since then, he has been dedicated to building Baidu's basic infrastructure. In 2013, he took full responsibility for Baidu Cloud's various operations, leading the construction and development of personal cloud, open cloud, cloud ecosystem, light applications, and Baidu smart hardware.

Yesterday, Teacher Zhenyu and Teacher Yu Kai discussed using images to establish connections between people and the world. It's cool! But how exactly do we connect? Let me reveal the secrets.

Light Tap: The New Way to Play

Teacher Zhenyu told us how to connect "people and services." He focused on "Light Tap," which is Baidu's new way to play in the mobile internet.

In Zhenyu's view, Light Tap connects people and services. Because he believes Light Tap is an entry point for the mobile internet : you can take a Light Tap photo of a movie poster, and what comes out might be ticket booking services; you can use Light Tap to photograph a plant, and what comes out might be common knowledge services provided by Baidu Baike. In short, Light Tap is the entry point that connects people with various services.

In my opinion, Light Tap is indeed promising because it makes accessing services extremely simple - I only need to take a photo to get a service.

Cameras Will Connect the World

At the Xierqi Night Talk现场, Teacher Yu Kai said that 90% of human cognition comes from vision , and vision consists of images and videos (videos are actually just a series of still images stitched together).

About visual perception, he also gave an example: the 2005 Vatican papal election, when there wasn't much photography happening; by 2013, almost everyone attending the Vatican papal election was using their phones to take photos. This shows that mobile cameras have become an extension of visual perception, and the imagination they bring is enormous.

"In the future, cameras will connect people and the world," Teacher Yu Kai said boldly.

Who Will Be the Connection Method for Mobile Internet?

Before the opening, the现场's 20+ audience members conducted a small survey called "big question" - "In O2O scenarios, which connection method do you think is most promising?" Among the four options of QR codes, image recognition, NFC, and voice recognition, image recognition won with 15 votes, sweeping the other options. NFC got 2 votes, voice recognition got 3 votes, while QR codes only got 1 vote.

Teachers Hou Zhenyu and Yu Kai both provided explanations for these different services. They believe voice recognition can very accurately express a person's needs and is also a way to connect services. However, voice recognition has limitations. Teacher Yu Kai said: "If you're shouting ads in an elevator, how awkward would that be?" In Yu Kai's view, QR codes are simple human-computer interaction, which is a different concept from connecting people and the world through cameras.

Hou Zhenyu believes that compared to other services, images are more comprehensive. Perhaps you can't describe the appearance of a beautiful woman, but you only need to use a camera to photograph her (of course, not secretly), then hand it over to image recognition to connect to a service, and the machine can tell you which star she is.

Playing in the Cloud: Openness is the "Hard Truth"

One thing that left a deep impression on me at the现场 was when Hou Zhenyu said, "Light Tap is just one of Baidu's products. Baidu Cloud platform wants to contribute the latest technology to the industry, and all our technologies are open to everyone. We are fully prepared for comprehensive openness."

This forms a sharp contrast with certain internet giants that block each other.

I'm a Java engineer, and I'm very aware that providing services in the mobile internet requires very powerful backend infrastructure. This cost is not something small and medium-sized mobile developers can bear. Now that Baidu is willing to provide Baidu Open Cloud to developers and help them build a cloud ecosystem, it's simply a blessing for entrepreneurs.

Of course, from another perspective, providing services in the mobile internet is based on cloud storage and big data. All this big data is accumulated from user data. If you work behind closed doors, this data will be very limited. Therefore, only openness can build a good cloud ecosystem and truly achieve connections between people and services.