test error

[Hopedreams]

Using the author's code,the mean angle error of the model is 12.8 degrees.There is a certain gap with the results in the essay.What could be the reason?

[RSKothari]

You are most likely reporting the validation result.

On Sat, Jun 26, 2021, 11:09 PM Hopedreams @.***> wrote:

Using the author's code,the mean angle error of the model is 12.8 degrees.There is a certain gap with the results in the essay.What could be the reason?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/erkil1452/gaze360/issues/28, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACPIEKVCUNNVCEYL2PPHBCDTU2I7FANCNFSM47L6OCVA .

[RSKothari]

Could you clarify which result you are referring to? Are you trying to reproduce the frontal results?

The code is designed to reproduces 13.5 degrees ( all 360 degree )

On Sat, Jun 26, 2021, 11:21 PM Hopedreams @.***> wrote:

[image: image] https://user-images.githubusercontent.com/81133186/123531715-7d495e00-d739-11eb-9fa1-d0fbf156c3b8.png Results in the essay is 11.4 degree. How can I solve this problem?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/erkil1452/gaze360/issues/28#issuecomment-869093671, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACPIEKWV7NVIZDSDRT5BR6TTU2KLJANCNFSM47L6OCVA .

[RSKothari]

Hmm .. I have reproduced the authors code many times and I can confirm that I can reproduce 13.5. I recommend first downloading the weights provided by the authors and confirm if you can reproduce 13.5. If not, then it indicates some data files might be missing.

[Hopedreams]

Thank you very much.Inspired by you,I reconfirmed and the result is 13.5.

[Hopedreams]

Why are pitch and yaw defined in this way?

[erkil1452]

The tanh should not be interpreted geometrically. It is a common nonlinearity used in machine learning and we use it to squeeze the output to the range (-1,1). This is then easy to rescale to the range required for yaw (-pi, pi) and pitch (-pi/2, pi/2) angles as seen above.

[Hopedreams]

Is the output of the model a three-dimensional vector?

------------------ 原始邮件 ------------------ 发件人: "erkil1452/gaze360" @.>; 发送时间: 2021年10月2日(星期六) 凌晨2:39 @.>; @.**@.>; 主题: Re: [erkil1452/gaze360] test error (#28)

The tanh should not be interpreted geometrically. It is a common nonlinearity used in machine learning and we use it to squeeze the output to the range (-1,1). This is then easy to rescale to the range required for yaw (-pi, pi) and pitch (-pi/2, pi/2) angles as seen above.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android.

[erkil1452]

The output is yaw and pitch in polar coordinates which can be converted into 3D direction using

x = sin(yaw) * cos(pitch)
y = sin(pitch)
z = cos(yaw) * cos(pitch)

[Hopedreams]

Does this 3D direction refer to the coordinates (x, y, z) of the vector in the Cartesian coordinate system?

[erkil1452]

Yes, it is a 3D direction in the Cartesian coordinate system of the camera.

[Hopedreams]

What do these three output values represent respectively? Thank you very much

------------------ 原始邮件 ------------------ 发件人: "erkil1452/gaze360" @.>; 发送时间: 2021年11月3日(星期三) 晚上10:32 @.>; @.**@.>; 主题: Re: [erkil1452/gaze360] test error (#28)

Yes, it is a 3D direction in the Cartesian coordinate system of the camera.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android.

[erkil1452]

The gaze vector components (x, y and z).