( for PAN CAM 3.0) it would be VERY cool if you could add AI control of the camera panning to follow the trigger/event, so that when it moves X amount of pixels it follows the angle of movement to keep the trigger in the center of view every time it moves, of course youll have to have a stepping of camera movement to be able to detect pixel movement, YOU CAN DO IIIIIT
They already did it. The PanCam 3.0 (PanV3) was released in January.
All PanCams, including the PanV3 already do this… To an extent.
All the PanCams have Step Motors. They all have Motion Tracking features that will Pan and Tilt the cam to follow the motion of an object when a Motion Event occurs.
All the PanCams also use pixelation algorithms to detect motion within the FOV. The degree of motion pixelation that will trigger a Motion Event and initiate Motion Tracking is determined by your Sensitivity settings.
Where the limitation lies is in the centering of the motion object in the FOV. But, this is by design. When the cam is panning and tilting, the quality of the image and the detection capability of the cam is severely limited because the cam is creating its own pixelation motion in the FOV by moving the cam. Motion in the FOV combined with motion from the cam creates a very low quality video image.
To minimize this, the cam needs to move as little as possible during a motion tracking event so that it can be stationary to capture high quality video. It has built in logic that detects when the object is leaving frame and pans or tilts to get ahead of it. If it maintained center, the cam would be in constant motion and the video would be one big blur.
I think they are asking for panning based on AI object detection vs general pixel-based motion. AI object panning would require a redesign (Cam Pan v4?) as it would require onboard AI processing vs the cloud-based AI processing of the Cam Pan v3. If the OP confirms this desire/intent, I’ll merge this request to the Pan v4 wishlist topic.
yes, but it should still be able to detect the pixels also, in case something else is detected, the thought was Once something moves a certain (X adj setting) amount AI follows the movement but only the object is moving ,not the camera ( AI is following), the camera will still trigger record but only move in steps ( when it is X adj distance from the frame edge) so pixelation only happens during when the camera needs to step ( adj setting for this too) \8O oooooh maybe you can set up a link with AI between camera’s( multi camera control) so a 2nd camera can take over from the 1rst. i had this thought because i have a perimeter wire free robot mower and wanted to use my camera to record it while working, my floodlight cam see’s most of my side yard, but not the back,( i have a pan cam there) and i didn’t get one for the front “yet” but WOW if AI was watching on the floodlight cam to enable a “ready follow object” or something the pan cam could intercept at the edge of the frame and not have to move the cam until the opposite side frame is reached. ught ohhh, what did i just do. DAVE…what are you doing dave, i can’t let you do that dave ( SO 2000) LOL, its ok hal, im just tweaking your video adjusters. … SORRY i had to
1 Like
All features you listed already exist in the wishlist category as separate topics, including linking tracking between multiple Cam Pans. The Can Pan v3 already attempts to keep motion object centered and already does this via motorized stepping on X/Y axes. What we (users) don’t have control of or know the current values of is granularity of step distance and speed, and multiple object prioritization.
What is not clear is your definition and use of “AI” as applied to the features you listed. All 3 current Cam Pan models do not have onboard AI processing… they detect motion only based on pixel change. I.e., pixel-based motion event is sent to cloud for AI detection, cloud returns AI actions back to client. If your intent is to have the Cam Pan v3 perform any of your features based on AI object detection (e.g., Person, Pet, Vehicle, etc.), the result would be useless due to client<->cloud server lag. To accomplish your goals using simple pixel-based motion is somewhat feasible and already requested, but it is not feasible via AI object recognition until Wyze develops a pan cam with onboard AI detection.
The ultimate solution to your use case would be to allow you (user) to define your new AI object (robot mower) in order for the cam to recognize the mower and place priority on this type of AI object for tracking. That AI feature is currently being developed and beta tested by Wyze as “Smart Vision”.
1 Like