search

bennycheung / Food100_YOLO_Tools

Python tools and configuration files for Food100 dataset DarkNet YOLO training
56 stars 22 forks source link

cannot load image #1

Closed NarenBabuR closed 6 years ago

NarenBabuR commented 6 years ago

I have done all steps as you said. But I cant load the images. It throws me an error

bennycheung commented 6 years ago

Can you show me the error? one reminder, this set of tools is going with the post http://bennycheung.github.io/yolo-for-real-time-food-detection It requires (1) Food100 dataset has been downloaded (2) food100_tk_label_bbox.py requires TKinter and PIL packages

NarenBabuR commented 6 years ago

Hello,

Thanks for reply... My problem got solved...

By the way your implementation was interesting

On Sun 10 Jun, 2018, 18:59 Benny Cheung, notifications@github.com wrote:

Can you show me the error? one reminder, this set of tools is going with the post http://bennycheung.github.io/yolo-for-real-time-food-detection It requires (1) Food100 dataset has been downloaded (2) food100_tk_label_bbox.py requires TKinter and PIL packages

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bennycheung/Food100_YOLO_Tools/issues/1#issuecomment-396049215, or mute the thread https://github.com/notifications/unsubscribe-auth/AhLxTT1DyR8Eo3xI-ks38HhNka_3LHKIks5t7R8fgaJpZM4Uhtqr .

bennycheung commented 6 years ago

Thanks, I am glad that your problem resolved. Enjoy!

NarenBabuR commented 6 years ago

Hai,

Can u please give me just an ML approach to solve the below problem

Scenario: There is a surveillance camera in a class.

Problem :(Important) Find the Machine Learning approach to find the malpractice in the class

We can't perform Supervised approach because we don't have enough data..

On Mon 11 Jun, 2018, 00:38 Benny Cheung, notifications@github.com wrote:

Closed #1 https://github.com/bennycheung/Food100_YOLO_Tools/issues/1.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bennycheung/Food100_YOLO_Tools/issues/1#event-1672739803, or mute the thread https://github.com/notifications/unsubscribe-auth/AhLxTdQhadKXcSsllnkC4pekgZ1Ye4Rsks5t7W6xgaJpZM4Uhtqr .

bennycheung commented 6 years ago

I see your problem is very interesting, to monitor malpractice in a classroom.

Unfortunately, with current deep learning technology, we need a large set of "label" data. In your case, images that identify certain behaviour in class is labelled as "malpractice". Also, with your problem, you are facing another challenge of bias data set, i.e. a lot of images of "normal" but only a few images of "malpractice".

I remembered in Kaggle. They have similar problem for "distracted driver detection": https://www.kaggle.com/c/state-farm-distracted-driver-detection

You may want to study this case for inspirations.

NarenBabuR commented 6 years ago

Thank you very much :))

On Fri 22 Jun, 2018, 18:29 Benny Cheung, notifications@github.com wrote:

I see your problem is very interesting, to monitor malpractice in a classroom.

Unfortunately, with current deep learning technology, we need a large set of "label" data. In your case, images that identify certain behaviour in class is labelled as "malpractice". Also, with your problem, you are facing another challenge of bias data set, i.e. a lot of images of "normal" but only a few images of "malpractice".

I remembered in Kaggle. They have similar problem for "distracted driver detection": https://www.kaggle.com/c/state-farm-distracted-driver-detection

You may want to study this case for inspirations https://www.youtube.com/watch?v=jgpP0ZbGGc8

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/bennycheung/Food100_YOLO_Tools/issues/1#issuecomment-399434258, or mute the thread https://github.com/notifications/unsubscribe-auth/AhLxTXbxVqeO1Hd8q4dmsGOSZvdRBUYBks5t_OpMgaJpZM4Uhtqr .