arunponnusamy / gender-detection-keras

Gender detection (from scratch) using deep learning with keras and cvlib.
http://arunponnusamy.com
MIT License
184 stars 68 forks source link

Web cam #3

Open aliraza1616 opened 6 years ago

aliraza1616 commented 6 years ago

HI. i have checked Your code and its working fine on still. When i ran the model of Webcam, it always displayed man and women as man. can you please provide some help?

aliraza1616 commented 6 years ago

also if we want it to count the faces, how can we do it?

arunponnusamy commented 6 years ago

Hello @aliraza1616 , In this repository I have demonstrated how to train a neural network from scratch (without using transfer learning / fine tuning pre-trained model weights) for a task like 'gender detection'. The accuracy of the detection will be low mostly due to

  1. very small training dataset and
  2. shallow / lightweight network architecture (in this case SmallerVGGNet).

    I am working on building more accurate gender detector. I will be adding it to cvlib in the next release. Keep on eye on the blog for new posts.

Regarding counting faces, you can just keep a counting variable inside the loop and keep incrementing them whenever a face is detected and the control comes inside the loop.

aliraza1616 commented 6 years ago

Hi. I am trying to add images to data sets fkr its complete training. Secondly i have to add counter too. Can you please add a line in code for me? I will be highly obliged if you can. As i am new to this field and i dont know much. Please if possible.

Get Outlook for Androidhttps://aka.ms/ghei36


From: Arun Ponnusamy notifications@github.com Sent: Saturday, October 13, 2018 1:09:18 PM To: arunponnusamy/gender-detection-keras Cc: aliraza1616; Mention Subject: Re: [arunponnusamy/gender-detection-keras] Web cam (#3)

Hello @aliraza1616https://github.com/aliraza1616 , In this repository I have demonstrated how to train a neural network from scratch (without using transfer learning / fine tuning pre-trained model weights) for a task like 'gender detection'. The accuracy of the detection will be low mostly due to

  1. very small training dataset and
  2. shallow / lightweight network architecture (in this case SmallerVGGNet).

I am working on building more accurate gender detector. I will be adding it to cvlibhttp://cvlib.net in the next release. Keep on eye on the bloghttp://arunponnusamy.com for new posts.

Regarding counting faces, you can just keep a counting variable inside the loophttps://github.com/arunponnusamy/gender-detection-keras/blob/cb922b07c8189a967fbd8b24d12cdf9dd5846b41/detect_gender_webcam.py#L85 and keep incrementing them whenever a face is detected and the control comes inside the loop.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/arunponnusamy/gender-detection-keras/issues/3#issuecomment-429521190, or mute the threadhttps://github.com/notifications/unsubscribe-auth/Ak2iaOvtnVMw_kAYKG-58LTkJBfLZp_Cks5ukZ-ugaJpZM4XZ7hO.

aliraza1616 commented 6 years ago

Hi,

Can i run any Video in Gender recognition code? if yes kindly tell how to?


From: Arun Ponnusamy notifications@github.com Sent: Saturday, October 13, 2018 1:09 PM To: arunponnusamy/gender-detection-keras Cc: aliraza1616; Mention Subject: Re: [arunponnusamy/gender-detection-keras] Web cam (#3)

Hello @aliraza1616https://github.com/aliraza1616 , In this repository I have demonstrated how to train a neural network from scratch (without using transfer learning / fine tuning pre-trained model weights) for a task like 'gender detection'. The accuracy of the detection will be low mostly due to

  1. very small training dataset and
  2. shallow / lightweight network architecture (in this case SmallerVGGNet).

I am working on building more accurate gender detector. I will be adding it to cvlibhttp://cvlib.net in the next release. Keep on eye on the bloghttp://arunponnusamy.com for new posts.

Regarding counting faces, you can just keep a counting variable inside the loophttps://github.com/arunponnusamy/gender-detection-keras/blob/cb922b07c8189a967fbd8b24d12cdf9dd5846b41/detect_gender_webcam.py#L85 and keep incrementing them whenever a face is detected and the control comes inside the loop.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/arunponnusamy/gender-detection-keras/issues/3#issuecomment-429521190, or mute the threadhttps://github.com/notifications/unsubscribe-auth/Ak2iaOvtnVMw_kAYKG-58LTkJBfLZp_Cks5ukZ-ugaJpZM4XZ7hO.

arunponnusamy commented 6 years ago

Hello @aliraza1616 , you can replace the zero in VideoCapture(0) with path to video file name to work with video input. VideoCapture('video_input.mp4') https://github.com/arunponnusamy/gender-detection-keras/blob/cb922b07c8189a967fbd8b24d12cdf9dd5846b41/detect_gender_webcam.py#L23

aliraza1616 commented 6 years ago

Hello. Let me check. Bro i have requested you for face count in the same code. How can we count faces and display in output screen? Plz support this case.

Get Outlook for Androidhttps://aka.ms/ghei36


From: Arun Ponnusamy notifications@github.com Sent: Saturday, October 20, 2018 10:35:05 AM To: arunponnusamy/gender-detection-keras Cc: aliraza1616; Mention Subject: Re: [arunponnusamy/gender-detection-keras] Web cam (#3)

Hello @aliraza1616https://github.com/aliraza1616 , you can replace the zero in VideoCapture(0) with path to video file name to work with video input. VideoCapture('video_input.mp4') https://github.com/arunponnusamy/gender-detection-keras/blob/cb922b07c8189a967fbd8b24d12cdf9dd5846b41/detect_gender_webcam.py#L23

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/arunponnusamy/gender-detection-keras/issues/3#issuecomment-431551147, or mute the threadhttps://github.com/notifications/unsubscribe-auth/Ak2iaBDm_wE9ZDoXWqu9eYbv5aY6OWfiks5umrYJgaJpZM4XZ7hO.

aliraza1616 commented 6 years ago

Hi!! I need t know that some of the faces are not detected by model. Like in attached pic. How to increase accuracy of detecting? Kindly reply for both the queries. Face count and face detection.

BR, Ali

Sent from Mailhttps://go.microsoft.com/fwlink/?LinkId=550986 for Windows 10


From: Arun Ponnusamy notifications@github.com Sent: Saturday, October 20, 2018 10:35:05 AM To: arunponnusamy/gender-detection-keras Cc: aliraza1616; Mention Subject: Re: [arunponnusamy/gender-detection-keras] Web cam (#3)

Hello @aliraza1616https://github.com/aliraza1616 , you can replace the zero in VideoCapture(0) with path to video file name to work with video input. VideoCapture('video_input.mp4') https://github.com/arunponnusamy/gender-detection-keras/blob/cb922b07c8189a967fbd8b24d12cdf9dd5846b41/detect_gender_webcam.py#L23

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/arunponnusamy/gender-detection-keras/issues/3#issuecomment-431551147, or mute the threadhttps://github.com/notifications/unsubscribe-auth/Ak2iaBDm_wE9ZDoXWqu9eYbv5aY6OWfiks5umrYJgaJpZM4XZ7hO.

arunponnusamy commented 6 years ago

Can you attach the image please ?

Regarding face count, as I have mentioned earlier, you can just keep a counting variable inside the loop and keep incrementing them whenever a face is detected and the control comes inside the loop. https://github.com/arunponnusamy/gender-detection-keras/blob/cb922b07c8189a967fbd8b24d12cdf9dd5846b41/detect_gender_webcam.py#L85

If you are new to Python, I would suggest to go through some introductory material on realpython or codecademy.

aliraza1616 commented 6 years ago

Image is already attached bro.

Get Outlook for Androidhttps://aka.ms/ghei36


From: Arun Ponnusamy notifications@github.com Sent: Monday, October 22, 2018 4:31:36 PM To: arunponnusamy/gender-detection-keras Cc: aliraza1616; Mention Subject: Re: [arunponnusamy/gender-detection-keras] Web cam (#3)

Can you attach the image please ?

Regarding face count, as I have mentioned earlier, you can just keep a counting variable inside the loop and keep incrementing them whenever a face is detected and the control comes inside the loop. https://github.com/arunponnusamy/gender-detection-keras/blob/cb922b07c8189a967fbd8b24d12cdf9dd5846b41/detect_gender_webcam.py#L85

If you are new to Python, I would suggest to go through some introductory material on realpythonhttp://realpython.com or codecademyhttp://codecademy.com.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/arunponnusamy/gender-detection-keras/issues/3#issuecomment-431807871, or mute the threadhttps://github.com/notifications/unsubscribe-auth/Ak2iaMEy9OabT8zzOei5kzkWo4vPSq6Jks5unayYgaJpZM4XZ7hO.

arunponnusamy commented 6 years ago

Can you visit the current issue on github and attach the image ? replying to email directly with image doesn't actually help.

nyck33 commented 5 years ago

@arunponnusamy thanks for a great contribution.
do you think it's worth the time to use your model for transfer learning or do you think I should just start with a deeper Resnet or VGG16?

arunponnusamy commented 5 years ago

Hey @nyck33, If you are using ResNet / VGG16 I would suggest to start with ImageNet weights for transfer learning.

nyck33 commented 5 years ago

@arunponnusamy Thanks for the reply and I'm very sorry for the bomb of questions.
That''s a great idea but I'm also looking at VGGFace right now to use as the 'backbone'?

arunponnusamy commented 5 years ago

I haven't worked much with VGGFace but I guess you can do that too.