about conv2d_cd code in tensorflow about cdcn HOT 24 CLOSED

@xuhangxuhang

Can you please tell me about the data pre-processing.
The OULU dataset have videos, but the source code required the images,
Even that there is #13 map_dir, can you please explain to me about it?

from cdcn.

@xuhangxuhang

Can you please tell me about the data pre-processing.
The OULU dataset have videos, but the source code required the images,
Even that there is #13 map_dir, can you please explain to me about it?

I also want to know,Did you solve this problem?

from cdcn.

from cdcn.

Use PRNet to get the depth map, And mcnnn for boundary box
…
On Mon 31 Aug, 2020, 10:40 AM xiaoroulian, @.***> wrote: @xuhangxuhang https://github.com/xuhangxuhang Can you please tell me about the data pre-processing. The OULU dataset have videos, but the source code required the images, Even that there is #13 <#13> map_dir, can you please explain to me about it? I also want to know,Did you solve this problem? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#22 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMM7B2ZRVN73TUWXVFEMUCTSDMER7ANCNFSM4ODV3DNQ .

Hello，thank you for your guidance,but i'm still confused.
I know the MTCNN but don't know how to generate '.dat' files.
Could you tell me how to generate the '.dat' files or share me a related GitHub website?
Thank you very much!

from cdcn.

from cdcn.

from cdcn.

You can also use .dat On Sun 29 Nov, 2020, 1:25 AM Punitha Valli, [email protected] wrote:
…
Hi, Instead of .txt you can use .obj Thank you On Sun 29 Nov, 2020, 1:21 AM boom9807, @.> wrote: > Use PRNet to get the depth map, And mcnnn for boundary box > … <#m_-8412065392521917396_m_-7482038873794626729_> > On Mon 31 Aug, 2020, 10:40 AM xiaoroulian, @.> wrote: @xuhangxuhang > https://github.com/xuhangxuhang https://github.com/xuhangxuhang Can > you please tell me about the data pre-processing. The OULU dataset have > videos, but the source code required the images, Even that there is #13 > <#13> <#13 > <#13>> map_dir, can you please > explain to me about it? I also want to know,Did you solve this problem? — > You are receiving this because you commented. Reply to this email directly, > view it on GitHub <#22 (comment) > <#22 (comment)>>, or > unsubscribe > https://github.com/notifications/unsubscribe-auth/AMM7B2ZRVN73TUWXVFEMUCTSDMER7ANCNFSM4ODV3DNQ > . > > Hello，thank you for your guidance,but i'm still confused. > I know the MTCNN but don't know how to generate '.dat' files. > Could you tell me how to generate the '.dat' files or share me a related > GitHub website? > Thank you very much! > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub > <#22 (comment)>, or > unsubscribe > https://github.com/notifications/unsubscribe-auth/AMM7B2276HSB3HD45FTYH2DSSEWRJANCNFSM4ODV3DNQ > . >

Hi,
I want to know how to generate the'. dat' files required by the code instead of which format of the files i can use.
I'm still confused about how to use MTCNN to generate '.dat' or '.obj' files.
Can I contack you via wechat or email？
Thank you very much

from cdcn.

from cdcn.

Hi, .dat is one of the file format which may be a plain text or binary . You can google it, it's very easy topic and easy to use ,.. instead of .txt you can use .dat in that place . So you will get a .dat file with the boundary box details . Thank you
…
On Mon 30 Nov, 2020, 2:46 PM boom9807, @.> wrote: You can also use .dat On Sun 29 Nov, 2020, 1:25 AM Punitha Valli, @. wrote: … <#m_3448670533241493752_> Hi, Instead of .txt you can use .obj Thank you On Sun 29 Nov, 2020, 1:21 AM boom9807, @.> wrote: > Use PRNet to get the depth map, And mcnnn for boundary box > … <#m_-8412065392521917396_m_-7482038873794626729_> > On Mon 31 Aug, 2020, 10:40 AM xiaoroulian, @.> wrote: @xuhangxuhang https://github.com/xuhangxuhang > https://github.com/xuhangxuhang https://github.com/xuhangxuhang Can > you please tell me about the data pre-processing. The OULU dataset have > videos, but the source code required the images, Even that there is #13 <#13> > <#13 <#13>> <#13 <#13> > <#13 <#13>>> map_dir, can you please > explain to me about it? I also want to know,Did you solve this problem? — > You are receiving this because you commented. Reply to this email directly, > view it on GitHub <#22 <#22> (comment) > <#22 (comment) <#22 (comment)>>>, or > unsubscribe > https://github.com/notifications/unsubscribe-auth/AMM7B2ZRVN73TUWXVFEMUCTSDMER7ANCNFSM4ODV3DNQ > . > > Hello，thank you for your guidance,but i'm still confused. > I know the MTCNN but don't know how to generate '.dat' files. > Could you tell me how to generate the '.dat' files or share me a related > GitHub website? > Thank you very much! > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub > <#22 (comment) <#22 (comment)>>, or > unsubscribe > https://github.com/notifications/unsubscribe-auth/AMM7B2276HSB3HD45FTYH2DSSEWRJANCNFSM4ODV3DNQ > . > Hi, I want to know how to generate the'. dat' files required by the code instead of which format of the files i can use. I'm still confused about how to use MTCNN to generate '.dat' or '.obj' files. Can I contack you via wechat or email？ Thank you very much — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#22 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMM7B26U7J32RDLH7FHRZBDSSM5TBANCNFSM4ODV3DNQ .

Hi，
You said use mtcnn to get bbox，could you tell me about more deteils?
Thank you

from cdcn.

from cdcn.

I think you can read the paper , it's very simple to undergo the further process
…
On Mon 30 Nov, 2020, 3:43 PM boom9807, @.> wrote: Hi, .dat is one of the file format which may be a plain text or binary . You can google it, it's very easy topic and easy to use ,.. instead of .txt you can use .dat in that place . So you will get a .dat file with the boundary box details . Thank you … <#m_1427279175925379468_> On Mon 30 Nov, 2020, 2:46 PM boom9807, @.> wrote: You can also use .dat On Sun 29 Nov, 2020, 1:25 AM Punitha Valli, @. wrote: … <#m_3448670533241493752_> Hi, Instead of .txt you can use .obj Thank you On Sun 29 Nov, 2020, 1:21 AM boom9807, @.> wrote: > Use PRNet to get the depth map, And mcnnn for boundary box > … <#m_-8412065392521917396_m_-7482038873794626729_> > On Mon 31 Aug, 2020, 10:40 AM xiaoroulian, @.*> wrote: @xuhangxuhang https://github.com/xuhangxuhang https://github.com/xuhangxuhang > https://github.com/xuhangxuhang https://github.com/xuhangxuhang Can > you please tell me about the data pre-processing. The OULU dataset have > videos, but the source code required the images, Even that there is #13 <#13> <#13 <#13>> > <#13 <#13> <#13 <#13>>> <#13 <#13> <#13 <#13>> > <#13 <#13> <#13 <#13>>>> map_dir, can you please > explain to me about it? I also want to know,Did you solve this problem? — > You are receiving this because you commented. Reply to this email directly, > view it on GitHub <#22 <#22> <#22 <#22>> (comment) > <#22 <#22> (comment) <#22 (comment) <#22 (comment)>>>>, or > unsubscribe > https://github.com/notifications/unsubscribe-auth/AMM7B2ZRVN73TUWXVFEMUCTSDMER7ANCNFSM4ODV3DNQ > . > > Hello，thank you for your guidance,but i'm still confused. > I know the MTCNN but don't know how to generate '.dat' files. > Could you tell me how to generate the '.dat' files or share me a related > GitHub website? > Thank you very much! > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub > <#22 <#22> (comment) <#22 (comment) <#22 (comment)>>>, or > unsubscribe > https://github.com/notifications/unsubscribe-auth/AMM7B2276HSB3HD45FTYH2DSSEWRJANCNFSM4ODV3DNQ > . > Hi, I want to know how to generate the'. dat' files required by the code instead of which format of the files i can use. I'm still confused about how to use MTCNN to generate '.dat' or '.obj' files. Can I contack you via wechat or email？ Thank you very much — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#22 (comment) <#22 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMM7B26U7J32RDLH7FHRZBDSSM5TBANCNFSM4ODV3DNQ . Hi， You said use mtcnn to get bbox，could you tell me about more deteils? Thank you — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#22 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMM7B23SV6QLSYQ3M2WQFGTSSNELVANCNFSM4ODV3DNQ .

Hi,
Sorry to disturb you.
I didn't find anything about MTCNN in the paper.
Can you tell me in detail how to get the file like ' 6_3_20_5_121_scene.dat ' from the video 'Train_files/1_1_01_1.avi '?
or share me a related MTCNN GitHub website?

Thank you so much.

from cdcn.

from cdcn.

can you share your mail id, or else your email ID
…
On Mon, 30 Nov 2020 at 16:56, boom9807 @.> wrote: I think you can read the paper , it's very simple to undergo the further process … <#m_-6656214226946969690_> On Mon 30 Nov, 2020, 3:43 PM boom9807, @.> wrote: Hi, .dat is one of the file format which may be a plain text or binary . You can google it, it's very easy topic and easy to use ,.. instead of .txt you can use .dat in that place . So you will get a .dat file with the boundary box details . Thank you … <#m_1427279175925379468_> On Mon 30 Nov, 2020, 2:46 PM boom9807, @.> wrote: You can also use .dat On Sun 29 Nov, 2020, 1:25 AM Punitha Valli, @. wrote: … <#m_3448670533241493752_> Hi, Instead of .txt you can use .obj Thank you On Sun 29 Nov, 2020, 1:21 AM boom9807, @.> wrote: > Use PRNet to get the depth map, And mcnnn for boundary box > … <#m_-8412065392521917396_m_-7482038873794626729_> > On Mon 31 Aug, 2020, 10:40 AM xiaoroulian, @.> wrote: @xuhangxuhang https://github.com/xuhangxuhang https://github.com/xuhangxuhang https://github.com/xuhangxuhang > https://github.com/xuhangxuhang https://github.com/xuhangxuhang Can > you please tell me about the data pre-processing. The OULU dataset have > videos, but the source code required the images, Even that there is #13 <#13> <#13 <#13>> <#13 <#13> <#13 <#13>>> > <#13 <#13> <#13 <#13>> <#13 <#13> <#13 <#13>>>> <#13 <#13> <#13 <#13>> <#13 <#13> <#13 <#13>>> > <#13 <#13> <#13 <#13>> <#13 <#13> <#13 <#13>>>>> map_dir, can you please > explain to me about it? I also want to know,Did you solve this problem? — > You are receiving this because you commented. Reply to this email directly, > view it on GitHub <#22 <#22> <#22 <#22>> <#22 <#22> <#22 <#22>>> (comment) > <#22 <#22> <#22 <#22>> (comment) <#22 <#22> (comment) <#22 (comment) <#22 (comment)>>>>>, or > unsubscribe > https://github.com/notifications/unsubscribe-auth/AMM7B2ZRVN73TUWXVFEMUCTSDMER7ANCNFSM4ODV3DNQ > . > > Hello，thank you for your guidance,but i'm still confused. > I know the MTCNN but don't know how to generate '.dat' files. > Could you tell me how to generate the '.dat' files or share me a related > GitHub website? > Thank you very much! > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub > <#22 <#22> <#22 <#22>> (comment) <#22 <#22> (comment) <#22 (comment) <#22 (comment)>>>>, or > unsubscribe > https://github.com/notifications/unsubscribe-auth/AMM7B2276HSB3HD45FTYH2DSSEWRJANCNFSM4ODV3DNQ > . > Hi, I want to know how to generate the'. dat' files required by the code instead of which format of the files i can use. I'm still confused about how to use MTCNN to generate '.dat' or '.obj' files. Can I contack you via wechat or email？ Thank you very much — You are receiving this because you commented. Reply to this email directly, view it on GitHub < #22 <#22> (comment) <#22 (comment) <#22 (comment)>>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMM7B26U7J32RDLH7FHRZBDSSM5TBANCNFSM4ODV3DNQ . Hi， You said use mtcnn to get bbox，could you tell me about more deteils? Thank you — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#22 (comment) <#22 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMM7B23SV6QLSYQ3M2WQFGTSSNELVANCNFSM4ODV3DNQ . Hi, Sorry to disturb you. I didn't find anything about MTCNN in the paper. Can you tell me in detail how to get the file like ' 6_3_20_5_121_scene.dat ' from the video 'Train_files/1_1_01_1.avi '? or share me a related MTCNN GitHub website? Thank you so much. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#22 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMM7B2ZB6V5NFO7PUHBMWVDSSNM2VANCNFSM4ODV3DNQ .

My email ID is [email protected]
Thank you for your help

from cdcn.

I am more curious about the content in the .dat file
Is it coordinates ? features? or path file?

from cdcn.

from cdcn.

It's the boundary box information about the face
…
On Tue 1 Dec, 2020, 12:55 AM 610821216, @.***> wrote: I am more curious about the content in the .dat file Is it coordinates ? features? or path file? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#22 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMM7B2YRTWIKY6ACTC52C63SSPE7NANCNFSM4ODV3DNQ .

Sorry, this is my first time using MTCNN
boundary box information about the face is （x, y, width, height）? Any other information?
Is the file naming method the image ID?
ex: image223 --> image223.dat?
Can I use txt instead of dat?

Both depth map and original image are used MTCNN generate boundary box information ?
train_image and Depth_image

from cdcn.

from cdcn.

Sure, you can use .txt formet also You can read a paper for mtcnn "Joint Face Detection and Alignment using Multi- task Cascaded Convolutional Network" and get more information about MTCNN THANK YOU
…
On Tue 1 Dec, 2020, 7:38 PM 610821216, @.> wrote: It's the boundary box information about the face … <#m_-8479692241029335515_m_-1357640567243200738_> On Tue 1 Dec, 2020, 12:55 AM 610821216, @.> wrote: I am more curious about the content in the .dat file Is it coordinates ? features? or path file? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#22 (comment) <#22 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMM7B2YRTWIKY6ACTC52C63SSPE7NANCNFSM4ODV3DNQ . Sorry, this is my first time using MTCNN boundary box information about the face is （x, y, width, height）? Any other information? Is the file naming method the image ID? ex: image223 --> image223.dat? Can I use txt instead of dat? Both depth map and original image are used MTCNN generate boundary box information ? train_image and Depth_image — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#22 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMM7B23V3PWTHTO5CXGU3XLSSTISVANCNFSM4ODV3DNQ .

Thank you for your helping

from cdcn.

Hi, @ZitongYu thanks for your codes, I really appreciate it.
I checked the Tensorflow version of CDC Conv op just now, I found that the definition of CDC Conv in TensorFlow implementation is not the same as the formula you proposed in your paper. You used
kernel_diff = tf.tile(kernel_diff, [kernel_size, kernel_size, 1, 1])
in the kernel_diff line, you copied the sum of kernel weights to a 3x3 weights filter, so the final result based on the TensorFlow version is not a 1x1 convolution on the current center of the receptive field. Can you explain why this happened?

(batch_size) x T x C X H X W，想问一下，这里的T是什么啊

from cdcn.

Sure, you can use .txt formet also You can read a paper for mtcnn "Joint Face Detection and Alignment using Multi- task Cascaded Convolutional Network" and get more information about MTCNN THANK YOU
…
On Tue 1 Dec, 2020, 7:38 PM 610821216, @.> wrote: It's the boundary box information about the face … <#m_-8479692241029335515_m_-1357640567243200738_> On Tue 1 Dec, 2020, 12:55 AM 610821216, @.> wrote: I am more curious about the content in the .dat file Is it coordinates ? features? or path file? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#22 (comment) <#22 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMM7B2YRTWIKY6ACTC52C63SSPE7NANCNFSM4ODV3DNQ . Sorry, this is my first time using MTCNN boundary box information about the face is （x, y, width, height）? Any other information? Is the file naming method the image ID? ex: image223 --> image223.dat? Can I use txt instead of dat? Both depth map and original image are used MTCNN generate boundary box information ? train_image and Depth_image — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#22 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMM7B23V3PWTHTO5CXGU3XLSSTISVANCNFSM4ODV3DNQ .
Hi,
Sorry to disturb you.
I didn't find anything about MTCNN in the paper.
Could you share me a related MTCNN GitHub website?
I'm also a student.
Thanks very much.

from cdcn.

Does anyone know how to implement this method in tensorflow version?

from cdcn.

from cdcn.

hmm.. I mean the Central Difference Convolution, how does it work on tensorflow?

from cdcn.

from cdcn.