Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

the attributes dataset to train #6

Open
zxczhkzxc opened this issue May 30, 2018 · 2 comments
Open

the attributes dataset to train #6

zxczhkzxc opened this issue May 30, 2018 · 2 comments

Comments

@zxczhkzxc
Copy link

thank you very much for your project,it's very help.but i have problem about the classification of attributes to ask for advice,class such as color,style,pattern;how can i do to finish this trough this project? Your help is really appreciated!

@abhishekrana
Copy link
Owner

You can add more heads(one for color, one for style and one for pattern) on the base model (currently there are 2 heads, classification and regression). [see model.py: line 44]
But this would require that you already have labels i.e. color, style and pattern for each image(or part of image) which you are feeding into the model. You would have to modify the loss function accordingly. Then train all these 5 heads together.
But i guess these things might be hard to learn for the network.

However, for color detection, you can just map the image(i.e. area inside bounding box) to color frequency and take the color having the highest frequency. (Assuming that major portion of the image is of the color you are interested in)

@zxczhkzxc
Copy link
Author

thank you for your answer.can we use DeepFashion Dataset for classification of color, style and pattern ?
i have't find the color attribute yet.for color detection,do you know some demo for application?i have a try to use opencv to detect the main color in an image,but it did not succeed

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants