Over-fit with single training Image, and detecting the flipped version of the image? [on hold]
$begingroup$
If I training a classifier to detect a dog vs not dog, and just use 1 image of dog, and multiple images of not dog objects to train and over-fit the model, then if I horizontal flip this same image of the dog and ask the model to predict, will it be able to detect it as dog?
If for training there are 1,00,000 images of non-Dog class and 1,00,000 images of dog class, with the exact same image repeated 1,00,000 times, would it now be able to detect the horizontally flipped image of the dog as Dog?
image-classification overfitting
asked yesterday
Jibin MathewJibin Mathew
1345
$endgroup$
put on hold as too broad by Esmailian, Kiritee Gak, Tasos, Mark.F, Stephen Rauch yesterday
Please edit the question to limit it to a specific problem with enough detail to identify an adequate answer. Avoid asking multiple distinct questions at once. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.
add a comment |
$begingroup$
If I training a classifier to detect a dog vs not dog, and just use 1 image of dog, and multiple images of not dog objects to train and over-fit the model, then if I horizontal flip this same image of the dog and ask the model to predict, will it be able to detect it as dog?
If for training there are 1,00,000 images of non-Dog class and 1,00,000 images of dog class, with the exact same image repeated 1,00,000 times, would it now be able to detect the horizontally flipped image of the dog as Dog?
image-classification overfitting
asked yesterday
Jibin MathewJibin Mathew
1345
$endgroup$
put on hold as too broad by Esmailian, Kiritee Gak, Tasos, Mark.F, Stephen Rauch yesterday
Please edit the question to limit it to a specific problem with enough detail to identify an adequate answer. Avoid asking multiple distinct questions at once. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.
add a comment |
$begingroup$
If I training a classifier to detect a dog vs not dog, and just use 1 image of dog, and multiple images of not dog objects to train and over-fit the model, then if I horizontal flip this same image of the dog and ask the model to predict, will it be able to detect it as dog?
If for training there are 1,00,000 images of non-Dog class and 1,00,000 images of dog class, with the exact same image repeated 1,00,000 times, would it now be able to detect the horizontally flipped image of the dog as Dog?
image-classification overfitting
asked yesterday
Jibin MathewJibin Mathew
1345
$endgroup$
If I training a classifier to detect a dog vs not dog, and just use 1 image of dog, and multiple images of not dog objects to train and over-fit the model, then if I horizontal flip this same image of the dog and ask the model to predict, will it be able to detect it as dog?
If for training there are 1,00,000 images of non-Dog class and 1,00,000 images of dog class, with the exact same image repeated 1,00,000 times, would it now be able to detect the horizontally flipped image of the dog as Dog?
image-classification overfitting
image-classification overfitting
asked yesterday
Jibin MathewJibin Mathew
1345
asked yesterday
Jibin MathewJibin Mathew
1345
edited 20 hours ago
Jibin Mathew
asked yesterday
Jibin MathewJibin Mathew
1345
asked yesterday
Jibin MathewJibin Mathew
1345
asked yesterday
Jibin MathewJibin Mathew
1345
1345
put on hold as too broad by Esmailian, Kiritee Gak, Tasos, Mark.F, Stephen Rauch yesterday
Please edit the question to limit it to a specific problem with enough detail to identify an adequate answer. Avoid asking multiple distinct questions at once. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.
put on hold as too broad by Esmailian, Kiritee Gak, Tasos, Mark.F, Stephen Rauch yesterday
Please edit the question to limit it to a specific problem with enough detail to identify an adequate answer. Avoid asking multiple distinct questions at once. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.
add a comment |
add a comment |
1 Answer
1
active
oldest
votes
$begingroup$
It should work with data augmentation. Without data augmentation, there are several implementation problems, such as :
- Class imbalance : If you have 10 other images, only 1 is dog.
Difficult to identify proper loss function for such extreme
imbalance. - Batch learning wont work (Only 1 batch will have
positive example)
answered yesterday
Shamit VermaShamit Verma
1,4291214
$endgroup$
$begingroup$
As I mentioned, there are only 2 classes, and model is already over-fit, my question if the flipped dog image would work.
$endgroup$
– Jibin Mathew
yesterday
$begingroup$
does it recognize non-dog images as dog image ? since it is over-fit, which class it did not recognize correctly (dog or non-dog)
$endgroup$
– Shamit Verma
yesterday
$begingroup$
Does it recognize the horizontally flipped image of dog as dog?
$endgroup$
– Jibin Mathew
21 hours ago
$begingroup$
Is this a hypothetical question, or have you already implemented the model ?
$endgroup$
– Shamit Verma
21 hours ago
$begingroup$
It will probably not even recognize dog due to class imbalance.
$endgroup$
– Shamit Verma
21 hours ago
|
show 2 more comments
1 Answer
1
active
oldest
votes
1 Answer
1
active
oldest
votes
active
oldest
votes
active
oldest
votes
$begingroup$
It should work with data augmentation. Without data augmentation, there are several implementation problems, such as :
- Class imbalance : If you have 10 other images, only 1 is dog.
Difficult to identify proper loss function for such extreme
imbalance. - Batch learning wont work (Only 1 batch will have
positive example)
answered yesterday
Shamit VermaShamit Verma
1,4291214
$endgroup$
$begingroup$
As I mentioned, there are only 2 classes, and model is already over-fit, my question if the flipped dog image would work.
$endgroup$
– Jibin Mathew
yesterday
$begingroup$
does it recognize non-dog images as dog image ? since it is over-fit, which class it did not recognize correctly (dog or non-dog)
$endgroup$
– Shamit Verma
yesterday
$begingroup$
Does it recognize the horizontally flipped image of dog as dog?
$endgroup$
– Jibin Mathew
21 hours ago
$begingroup$
Is this a hypothetical question, or have you already implemented the model ?
$endgroup$
– Shamit Verma
21 hours ago
$begingroup$
It will probably not even recognize dog due to class imbalance.
$endgroup$
– Shamit Verma
21 hours ago
|
show 2 more comments
$begingroup$
It should work with data augmentation. Without data augmentation, there are several implementation problems, such as :
- Class imbalance : If you have 10 other images, only 1 is dog.
Difficult to identify proper loss function for such extreme
imbalance. - Batch learning wont work (Only 1 batch will have
positive example)
answered yesterday
Shamit VermaShamit Verma
1,4291214
$endgroup$
$begingroup$
As I mentioned, there are only 2 classes, and model is already over-fit, my question if the flipped dog image would work.
$endgroup$
– Jibin Mathew
yesterday
$begingroup$
does it recognize non-dog images as dog image ? since it is over-fit, which class it did not recognize correctly (dog or non-dog)
$endgroup$
– Shamit Verma
yesterday
$begingroup$
Does it recognize the horizontally flipped image of dog as dog?
$endgroup$
– Jibin Mathew
21 hours ago
$begingroup$
Is this a hypothetical question, or have you already implemented the model ?
$endgroup$
– Shamit Verma
21 hours ago
$begingroup$
It will probably not even recognize dog due to class imbalance.
$endgroup$
– Shamit Verma
21 hours ago
|
show 2 more comments
$begingroup$
It should work with data augmentation. Without data augmentation, there are several implementation problems, such as :
- Class imbalance : If you have 10 other images, only 1 is dog.
Difficult to identify proper loss function for such extreme
imbalance. - Batch learning wont work (Only 1 batch will have
positive example)
answered yesterday
Shamit VermaShamit Verma
1,4291214
$endgroup$
It should work with data augmentation. Without data augmentation, there are several implementation problems, such as :
- Class imbalance : If you have 10 other images, only 1 is dog.
Difficult to identify proper loss function for such extreme
imbalance. - Batch learning wont work (Only 1 batch will have
positive example)
answered yesterday
Shamit VermaShamit Verma
1,4291214
answered yesterday
Shamit VermaShamit Verma
1,4291214
answered yesterday
Shamit VermaShamit Verma
1,4291214
answered yesterday
Shamit VermaShamit Verma
1,4291214
1,4291214
$begingroup$
As I mentioned, there are only 2 classes, and model is already over-fit, my question if the flipped dog image would work.
$endgroup$
– Jibin Mathew
yesterday
$begingroup$
does it recognize non-dog images as dog image ? since it is over-fit, which class it did not recognize correctly (dog or non-dog)
$endgroup$
– Shamit Verma
yesterday
$begingroup$
Does it recognize the horizontally flipped image of dog as dog?
$endgroup$
– Jibin Mathew
21 hours ago
$begingroup$
Is this a hypothetical question, or have you already implemented the model ?
$endgroup$
– Shamit Verma
21 hours ago
$begingroup$
It will probably not even recognize dog due to class imbalance.
$endgroup$
– Shamit Verma
21 hours ago
|
show 2 more comments
$begingroup$
As I mentioned, there are only 2 classes, and model is already over-fit, my question if the flipped dog image would work.
$endgroup$
– Jibin Mathew
yesterday
$begingroup$
does it recognize non-dog images as dog image ? since it is over-fit, which class it did not recognize correctly (dog or non-dog)
$endgroup$
– Shamit Verma
yesterday
$begingroup$
Does it recognize the horizontally flipped image of dog as dog?
$endgroup$
– Jibin Mathew
21 hours ago
$begingroup$
Is this a hypothetical question, or have you already implemented the model ?
$endgroup$
– Shamit Verma
21 hours ago
$begingroup$
It will probably not even recognize dog due to class imbalance.
$endgroup$
– Shamit Verma
21 hours ago
$begingroup$
As I mentioned, there are only 2 classes, and model is already over-fit, my question if the flipped dog image would work.
$endgroup$
– Jibin Mathew
yesterday
$begingroup$
As I mentioned, there are only 2 classes, and model is already over-fit, my question if the flipped dog image would work.
$endgroup$
– Jibin Mathew
yesterday
$begingroup$
does it recognize non-dog images as dog image ? since it is over-fit, which class it did not recognize correctly (dog or non-dog)
$endgroup$
– Shamit Verma
yesterday
$begingroup$
does it recognize non-dog images as dog image ? since it is over-fit, which class it did not recognize correctly (dog or non-dog)
$endgroup$
– Shamit Verma
yesterday
$begingroup$
Does it recognize the horizontally flipped image of dog as dog?
$endgroup$
– Jibin Mathew
21 hours ago
$begingroup$
Does it recognize the horizontally flipped image of dog as dog?
$endgroup$
– Jibin Mathew
21 hours ago
$begingroup$
Is this a hypothetical question, or have you already implemented the model ?
$endgroup$
– Shamit Verma
21 hours ago
$begingroup$
Is this a hypothetical question, or have you already implemented the model ?
$endgroup$
– Shamit Verma
21 hours ago
$begingroup$
It will probably not even recognize dog due to class imbalance.
$endgroup$
– Shamit Verma
21 hours ago
$begingroup$
It will probably not even recognize dog due to class imbalance.
$endgroup$
– Shamit Verma
21 hours ago
|
show 2 more comments
