We have an application that uses Google's ActivityRecognitionAPI and it works well for determining if a driver is driving, stopped, or when he hops out of vehicle and transitions to ON_FOOT.
我们有一个使用Google的ActivityRecognitionAPI的应用程序,它可以很好地确定驱动程序是否正在驾驶,停止或何时跳出车辆并转换为ON_FOOT。
However, we see a lot of false positives for IN_VEHICLE. With higher end devices it is much worse. We see them when users are typing emails or texts or even just navigating through device content.
但是,我们发现IN_VEHICLE有很多误报。使用更高端的设备会更糟糕。当用户输入电子邮件或文本或甚至只是浏览设备内容时,我们会看到它们。
Are we unique in seeing this? Do others have strategies for defending the false positives?
我们看到这个是独一无二的吗?其他人是否有策略来捍卫误报?
1 个解决方案
#1
0
This may come a little late but I have noticed that as well. I suppose it is because sometimes the way you move just looks like a vehicle's movement. I have noticed that sometimes I have similar false positives with on_bicycle. I have also noticed how often this happens because I am holding the phone in my hand and assuming an unusual position (e.g. I am retrieving my wallet from the pocket while walking).
这可能会有点晚,但我也注意到了。我想这是因为有时你移动的方式看起来像车辆的运动。我注意到有时候我和on_bicycle有类似的误报。我也注意到这种情况经常发生,因为我手里拿着电话并假设一个不寻常的位置(例如我在行走时从口袋里取回钱包)。
#1
0
This may come a little late but I have noticed that as well. I suppose it is because sometimes the way you move just looks like a vehicle's movement. I have noticed that sometimes I have similar false positives with on_bicycle. I have also noticed how often this happens because I am holding the phone in my hand and assuming an unusual position (e.g. I am retrieving my wallet from the pocket while walking).
这可能会有点晚,但我也注意到了。我想这是因为有时你移动的方式看起来像车辆的运动。我注意到有时候我和on_bicycle有类似的误报。我也注意到这种情况经常发生,因为我手里拿着电话并假设一个不寻常的位置(例如我在行走时从口袋里取回钱包)。