PTZ Camera Review Vol. 3 – Tracking and AI

• Affordable AVITOK “V449E2-NDI”

AVITOK is a video conferencing solution manufacturer headquartered in Shenzhen, Guangdong Province, China. Here, for the PTZ Camera Review, we picked up the 4K model PTZ camera “V449E2-NDI.” This model is equipped with AI person detection and verified the actual device. The estimated price is roughly $1100.

The V449E2-NDI is equipped with a Sony CMOS sensor that supports 4K UHD resolution, with resolutions up to 4K, frame rates up to 60fps, and 25X optical zoom. For human detection, it has a built-in processor, image processing, and analysis algorithms, and real-time tracking mode and zone tracking mode are available.

The user interface uses Web UI. When typing the IP address into your browser, an interface will appear where you can configure things like AI person detection. Since AVITOK is equipped with AI person detection as standard, its function settings are also included on one screen. Basically, you launch it on the Web UI, turn on person detection, stand in front of the camera, and it will automatically follow you. It’s simple, so you can use it without hesitation without having to read the manual.

Two AI Detection Modes

There are two types of AI person detection: “Presenter mode” and “Zone mode.” Presenter mode corresponds to a tracking function commonly referred to as AI person detection. Zone mode is a mode that is assigned to each scene. Images are lined up at the bottom of the screen, and up to four zones can be registered. For example, if you go to the left end of the second angle from the left, it will be at the same angle as the leftmost image.

This “Zone Mode” is most suitable for broadcasting schools and classes. You can also set angles such as “When you are on the left side of the blackboard, only the left side is shown” or “When you are on the right side of the blackboard, only the right side is shown.”

It does not support automatic framing by moving closer to or moving away from the camera, but I don’t think this will be a problem since the teacher does not move significantly back and forth during class delivery. If you want to choose a remote camera equipped with AI human detection, this seems to be the most likely candidate remote camera.

App Functions: Canon “CR-N700” + “Automatic tracking application”

• ・CR-N700 : Current market price: Approximately $8000 including tax
• ・Automatic tracking application RA-AT001 : Market price: Approximately $1300

Canon’s CR-N700 had a thoughtful design from the moment I set it up. In fact, the Web UI that each company has adopted often ends up wasting time because the IP address cannot be determined even when the user wants to start it. The CR-N700 also uses a web UI, but you can check the camera’s IP address using the “camera search tool.” Even if you don’t have any network knowledge, you won’t get lost as long as you are in the same network space. I thought Canon was very considerate in this regard.

　 　　

App Required for Auto-tracking

The CR-N700 does not come with an automatic tracking function as standard, but it can be used by purchasing and installing an automatic tracking app. Additional steps may be possible without reading the manual. I wasn’t particularly confused when testing these out.

Setting up the automatic tracking application is done using a six-step “settings wizard.” Users can set the silhouette size, tracking speed, background area, etc., and you can also set this without reading the manual.

Operating the App

Among the automatic tracking applications, what I thought was particularly excellent was the “Set composition” function. A silhouette is displayed on the screen, and you can set the display position of the tracking target by dragging the silhouette, and you can also set the amount of space above the person’s head using a sensory value. If you set the silhouette to the center or to the side, the actual image will move in real time. I thought this was quite innovative.

Another thing I liked is the setting when the tracking target is lost. When a tracking target is lost, many remote cameras return it to the front home position, but automatic tracking applications allow you to choose between “maintaining the angle of view” or “returning to the initial position” at the lost position.

Also, the return position can be set to the left or right instead of the front. For example, in the case of seminars, the speakers are often only in the upper right corner, so it is not ideal for the home position to be in the center. If the podium is on the right, it is also possible to set the home position based on that podium.

Subject Tracking: Sony “SRG-A40”

• ・SRG-A40 : Market price approximately $2900

Lastly, let’s introduce Sony’s “SRG-A40”. Sony calls the automatic tracking function “auto-framing.” It is called auto-framing because it not only has an automatic tracking function, but also allows you to frame the subject at a viewing angle that is comfortable for the viewer.

The camera is loaded with AI functions, and an AI graphics chip is mounted on the base, which is slightly thicker than a normal remote camera. This chip makes it possible to detect the human skeleton and the position of the head in images. By detecting the skeleton, the pose and orientation of the subject can also be determined, enabling highly accurate tracking.

The interface is Web UI. The SRG-A40 is equipped with automatic person detection as standard, and the function is built into the same page of the UI.

Auto-framing Functionality

Auto-framing settings are the same as Canon’s, and are done using a wizard. You can configure settings such as how much space above the subject is opened, the size of the person, and whether the person moves to the left or right. A feature of the SRG-A40 is that you can set the composition using pictograms, which allows you to set the composition without having someone stand in front of the camera.

In actual use, the SRG-A40’s human detection is more powerful than any other manufacturer’s. It can recognize even small profile faces, It will capture the selected person even when their back is turned. Also, it will hold track even when people cross paths. This feature is better than any other camera.

Although we do not have an example, it was possible to recognize it even when running vigorously. What was particularly surprising was that it was able to recognize even rough illustrations of people. Out of all these cameras, this has the best person-recognition abilities.

In terms of framing, the zooming of “people approaching” and “people moving back” is smooth, and shots of people walking sideways are captured in the center of the video. In this regard, I felt that he had the ability to create a natural and pleasant image.

PTZ Camera Review Vol. 3 Wrapup

All three models have completely different prices and features, so choose one depending on your intended use.

If you want to go all out and use it as an extension of your previous shooting, we recommend Canon’s remote camera + automatic tracking application. It may also be possible to replace the traditional use of professional photographers in professional settings with automatic tracking applications.

Sony’s SRG-A40 is suitable for fast-moving sports, dance, and stage shoots. Basically, it is almighty and can be used in any situation.

AVITOK’s V449E2-NDI is suitable for beginners at schools. Companies who have not used remote cameras but would like to introduce tracking functionality would like this too.