GrandpaXun242 / AdaBLDM

The implement for paper : "A Novel Approach to Industrial Defect Generation through Blended Latent Diffusion Model with Online Adaptation"
https://arxiv.org/abs/2402.19330
58 stars 1 forks source link

Training and Inference on AdaBLDM with Binary Masks Instead of Triple Masks #20

Open shierxtQ opened 2 months ago

shierxtQ commented 2 months ago

Hello,

I am working on training AdaBLDM with my custom dataset, which uses binary masks rather than the triple masks used in the original implementation. I would like to understand:

  1. Which parts of the code should be modified to adapt the model for binary masks during training and inference?
  2. Are there any specific settings or architectural changes needed to ensure compatibility and effective performance with binary masks?

I’ve looked through the documentation and codebase but couldn’t find detailed instructions on modifying the model for different mask formats. Any guidance or advice would be greatly appreciated!

GrandpaXun242 commented 2 months ago
  1. If you wanna to train a model with binary mask , you can change this lines for generating dateset with binary mask.
  2. I am a student without enough GPUs. I didn't try to train the model in all categories which spent too much times. You can try in training the model for single category like hazelnut.
GrandpaXun242 commented 2 months ago

Also , you should redesign the text prompt.