When the edge-side AI empowers intelligent imaging devices, this company sells millions of units globally in a year.

At 8 p.m. on October 29, the guests at a tech product launch event attracted industry attention.

The launch event invited 30 special guests - they were 30 mobile phones of various major brands. Different shooting software was enabled on them, and they were mounted on multiple Hohem AI stabilizers. When the host made an OK gesture, the 30 mobile phones followed the host's position for tracking and shooting in an orderly manner.

When a calligraphy enthusiast is immersed in writing, without the intervention of others, through the shooting target box selection on the touchscreen, the stabilizer can focus and lock on the enthusiast's hand, and follow the hand movements for shooting. At the same time, the controller screen on the handle can be separated and placed on the desktop to preview the shooting screen in real time, allowing one person to complete the entire shooting process immersively.

The stabilizer devices on which these "mobile phone audiences" are mounted come from Hohem, a tech company in Shenzhen that focuses on intelligent imaging equipment. This time, the "Hohem M7" mobile phone stabilizer they released is very advanced in the industry, both in technology and functionality.

In terms of design, the "Hohem M7" is not only an integrated form, but also adopts an ingenious modular design. Based on the orthogonal three-axis structure, it innovatively carries an AI tracking module and a touchscreen controller, and is equipped with a built-in high-intensity extension rod and full-color RGB light effects, which can be described as "fully stacked with features".

In terms of functionality, the "Hohem M7" not only has a high-performance three-axis anti-shake function, but also the "AI tracking" is comprehensively upgraded to the native follow-up shooting from tracking people to everything. It can also preview the AI screen in real time through the screen and remotely control the parameter adjustment.

Behind this function, it is precisely the insight into the demands of more segmented groups under the upsurge of short videos for all - They can be professional car shooting or photography geeks of landmark buildings, or users who shoot cute pets, or even people who don't want to show their faces and love handicrafts and art records.

Before the M7, why no brand could solve the problem of the mobile phone's "native follow-up shooting of everything"? In addition to having the vision of crowd insight, both technology and ideas are difficult problems.

For other brands to achieve object tracking, users need to download a specific APP or limit the mobile phone system to achieve native follow-up shooting. But as long as users must accommodate the machine, it is not a good enough intelligent product.

AI object tracking involves the model training of a large number of objects. Hohem adopts the technical idea of classifying object models and establishes a diverse sample image database of cars, pets, buildings, etc. But even with the foundation, the implementation still faces difficulties, because these images must be accurately recognized, transmitted to the controller screen with low latency, and the power consumption needs to be balanced, as well as the form of human-computer interaction needs to be planned.看似小小的模块，在背后是一套完整复杂的系统。The founder of Hohem, Chen Min, introduced that in order to overcome the problems of arbitrary object recognition, tracking and interaction experience, Hohem continuously ensures that the R & D team focuses and invests heavily, and finally integrates the three functions of AI screen preview, visual parameter adjustment, and remote control on the touchscreen controller of the Hohem M7 in a disruptive manner. The high integration and innovative interaction are indeed advanced in the industry.

It can be seen that what AI empowers the stabilizer is not the superposition of functions, but the redefinition of the operation mode and interaction experience.

After the release of the Hohem M7, it became the number one sales ranking in the Douyin 3C digital category within 2 hours.

This mobile phone stabilizer that combines anti-shake and stabilization, AI tracking, remote touch control, and intelligent camera movement is enough to prove that even in a vertical track, as long as continuous technological innovation is carried out to solve the actual pain points of segmented users, it can also quickly break through the circle in the current fiercely competitive intelligent imaging track.

It is precisely by continuously deepening in the field of stabilizers that its parent company, Hohem Excellence Technology, has also achieved considerable commercial success. In the past ten years, Hohem has emerged from the fierce hardware hotbed in Shenzhen and has grown into a national-level specialized, refined, special, and new "little giant" enterprise.

Today, the development of this company is just as its name suggests - Its annual revenue reaches hundreds of millions, and the shipment volume exceeds one million units and continues to rise. In overseas channels, their handheld stabilizer products have become the top category on platforms such as Amazon; in China, they have consecutively ranked first in the stabilizer category sales list on domestic platforms such as Douyin and Tmall.

The case of this typical Shenzhen company provides a fact: Although the penetration rate of handheld stabilizers is less than 1% at present, the professional vertical track is never a predicament, but contains huge opportunities.

The "Hohem M7" mobile phone stabilizer is a small observation window. Looking at the product ideas of Hohem over the past ten years, it can be found that the secret to the company's breakthrough lies in continuously promoting the intelligence and ease of use of intelligent imaging products.

According to Chen Min, the founder of Hohem, Hohem's investment in the AI algorithm model can be traced back to 2014. In fact, Hohem was the first manufacturer to empower the stabilizer with Bluetooth technology, allowing the intelligent tracking of the mobile phone stabilizer to be completed with the computing power and battery of the mobile phone. And as early as before this round of AI hardware wave this year, in 2021, Hohem has already breakthroughly empowered the "edge-side AI" capability to the V2 mobile phone stabilizer, further giving the mobile phone stabilizer the possibility of becoming an "intelligent photographer". Later, Hohem launched the new generation V3 mobile phone stabilizer, an innovative magnetic AI tracker that can freely switch the front/rear camera, and the tracking accuracy is greatly improved. The Hohem V3 has also just won the 2025 CES Innovation Award, a global top consumer technology industry benchmark.

When the Hohem stabilizer has been freed from the App limitation and can freely track the character target, the Hohem M7 has gone a step further, not limited to the character target, and has realized the recognition and tracking shooting of any object target in more scenarios. Obviously, only when the difficulty of using the hardware is gradually reduced, more intelligent and easy to use, can the product have the basis for universalization.

In the fierce business competition, few brands can maintain such a continuous enthusiasm for technological breakthroughs.

In the past ten years, Hohem has not relied on a hit product to conquer the world. While polishing the product, it also contains their extreme insight into the needs of different segmented groups in different scenarios, and focuses on penetrating the needs of scenario groups with a richer product matrix. For example, the product of the Hohem mobile phone stabilizer is further subdivided into different matrices - the M series corresponds to professional-level serious creative users, and the X series and V series correspond to consumer-level daily life recording users, so as to achieve a more refined marketing strategy from people, goods, and venues.

"Continuously carrying out product innovation to make AI follow-up stabilizers benefit more users is an important task for our future," said Chen Min, the founder and CEO of "Hohem". In his opinion, in the future, the stabilizer driven by edge-side AI still has greater possibilities to break through the circle, and this track is still very "vast" in the future.

Recently, in the communication with Chen Min by 36Kr, he shared how "Hohem" continuously innovates the handheld stabilizer technology, the iteration of the key product ideas behind it; and as a typical Shenzhen hard technology company, their determination to go global and break through the circle.

The following is the communication record (slightly edited)

When AI Becomes the Background of the Stabilizer, Is It a Plus or a Gimmick?

Q: For a category with a currently low population penetration rate, breaking through the circle is a very important thing. What are your breakthrough strategies and attempts?

Chen Min: The common saying we often talk about internally is: The product is small, portable, easy to use, and affordable. Our underlying logic is to move this product from the original niche professional photography equipment crowd to the general public.

Last year, we targeted the delicate mother group in China's Douyin. They need to record their daily lives more frequently. We specially developed a small and portable product and adopted a pink ID design language. We targeted and launched it on Douyin and sold more than 200,000 units. Before we reached this group, they actually had no understanding of the handheld stabilizer product category at all.

When we target the general consumer group, we focus on providing a better product interaction experience targeted at the characteristics of different groups and scenarios.

Q: What are the core AI technical difficulties that need to be overcome on the "Hohem M7"?

Chen Min: First, the AI algorithm. For common shooting subjects such as people, cars, and pets, the AI tracking conducts classified model training specifically, and the recognition accuracy and tracking effect are good. For shooting irregular object objects, the recognition success rate may be lower. For example, the recognition rate of a 45-degree car head may be high, but a slight deviation in the angle may result in inaccurate recognition. The Hohem team is also continuously optimizing for these scenarios.

Secondly, the handheld stabilizer is a battery-powered product. For edge-side AI, the core challenge is how to achieve a balance between low power consumption and AI computing power performance. We are continuously optimizing our model, such as gesture recognition, people, clothing, color blocks, and adaptive tracking of irregular shapes at different angles. With the further iteration of our algorithm, in the same computing power and power consumption situation, more AI shooting function experiences will be opened on our intelligent shooting system in the next one or two years.

Q: This time, the Hohem M7 claims to have two major features. Specifically, where is it strong?

Chen Min: One is the once again comprehensively upgraded AI tracker. Despite its appearance as just a magnetic camera, it integrates three modules: AI algorithm, image recognition, and fill light. It can not only achieve character/object tracking, but also the edge-side AI allows users to start tracking without downloading a specific APP, without connecting Bluetooth, and without limiting the mobile phone. Moreover, the fill light integrates two light modes: CCT and RGB full color. It is like installing a strong heart for the Hohem M7 and empowering the stabilizer.

The other is the industry's first split-type touchscreen controller. It integrates three functions: screen preview, visual parameter adjustment, and remote control. Our customized 1.4-inch full-color screen, combined with the real-time screen preview of AI tracking, and the magnetic split design, can be controlled even in long-distance scenes, saying goodbye to blind shooting.

Q: "Intelligent follow-up photographer", is it a pain point for users of a stabilizer, or is it an added bonus?

Chen Min: For users who record their daily lives, the underlying shooting needs are not fully released. The difficulty of shooting and editing videos makes everyone not fully enjoy the fun of recording and sharing in this scenario.

In the future, we hope to further empower through edge-side AI, so that many people who originally couldn't shoot can complete the recording and sharing tasks just by buying an intelligent shooting device.

Q: As mobile phone manufacturers are gradually improving the anti-shake function, will it have a dimensionality reduction blow to your industry?

Chen Min: Before 2021, the main function of the handheld stabilizer industry was only three-axis anti-shake, which is a pure execution unit. But after 2021, we gradually integrated the computing power of the edge-side into the handheld stabilizer, with AI visual perception ability, meaning that the stabilizer has iterated from the original pure execution unit to a small robot model with visual perception and execution closed-loop. On this basis, the stabilizer is more like the role of an "intelligent photographer", and anti-shake is just its most basic function.

Q: The functions of the Hohem stabilizer have also changed a lot in a few years. What is the iteration of the product idea behind this, and what are the key technological innovations?

Chen Min: The first stage: From electronic anti-shake to physical anti-shake. Like the electronic anti-shake of the mobile phone camera, it achieves the effect of stabilizing the image by cropping the picture, which is detrimental to the image quality. And electronic anti-shake is difficult to deal with large-scale shaking of people, so there is the physical anti-shake of the stabilizer. Users no longer need to worry about hand tremors and image quality damage.

The second stage: From anti-shake to the first-generation AI tracking. Before 2016, the stabilizer only had the anti-shake function. Hohem found that only "stable shooting" is too thin for users, and they also want the intelligent interaction of "help me shoot". Hohem was the first manufacturer to add Bluetooth APP and face tracking to the stabilizer. We launched the "intelligent shooting" product direction in 2016, which is the first-generation AI tracking.

The third stage: From the first-generation APP tracking to the edge-side AI tracking. After the APP tracking shooting was launched for a period of time, we saw some disadvantages - the user's pre-preparation work is relatively long, and all functions are completed in a specific APP. Therefore, in 2021, we were the first in the industry to integrate the edge-side computing power into the handheld stabilizer. The Hohem V2 mobile phone stabilizer product we launched does not rely on the mobile phone APP and Bluetooth connection. Just make an "OK" gesture to the machine, and the machine will quickly recognize this shooting target to achieve the effect of follow-up shooting in any shooting software.

The fourth stage: Further iteration of edge-side AI tracking. That is the recently newly released M7 by Hohem. The cause was that a user wearing a bamboo hat and dancing feedback that the tracking was not sensitive during the practice of a certain dance. We found that the user's clothing when dancing was very distinctive, and the thin veil hanging around the bamboo hat covered the human-shaped features. In our AI's visual recognition, it is a non-human feature, and in the algorithm logic at that time, it was not recognized as a human target. Starting from this user feedback, we further investigated and observed that many immersive live-streaming and handicraft users may not want to show their faces, but only want to shoot a certain body part. Therefore, the Hohem M7 has achieved tracking from people to everything. At the same time, the Hohem M7 also innovatively carries a touchscreen, allowing the AI screen to be previewed in real time through the screen. In addition to the original gesture interaction, it can also be framed and operated on the touchscreen, making