InstructDiffusion

InstructDiffusion copied to clipboard

zhuzhuzhu001

Thanks for your great work. I notice that you use COCO, CrowdPose, MPII, and AIC datasets for keypoint estimation. Have you ablated the influence of the diversity of the dataset...

SxJyJay

No interface is running right now

huangjch526

Hi, I see that in ./stable_diffusion/ldm/models/diffusion/ddpm_edit.py, params uncond=0.05 in `def get_input( )` is not used, replaced with fixed 0.075:  Also, bug: @torch.no_grad() has been commented out.

Sutongtong233

prompt about classification and detection

3

Thanks for your great work! I am curious about the the result of classification and detection in your paper:  You mention that you design special prompts for...

wenqsun

The repo contains a broken link to "Web demo", leading to page with the text "No interface is running right now".

sedol1339

Supported training phase

8

Thanks for your great work! I notice that your [config file](https://github.com/cientgu/InstructDiffusion/blob/main/configs/instruct_diffusion.yaml) load the checkpoint "v1-5-pruned-emaonly-adaption.ckpt", which seems correspond to the "Pretraining adaptation" phase in your main paper. Meanwhile, it seems...

SxJyJay

Checkpoint for Training

3

Hi! Thanks for your impressive work! I am trying to run the task-specific training and need to load the checkpoint generated in the [adaptation phase](https://github.com/cientgu/InstructDiffusion/blob/main/configs/instruct_diffusion.yaml#L11). Could you please provide this...

wusize

Questions about the post-processing for keypoint detection and segmentation

2

Hi, very interesting work! As mentioned in the paper, a lightweight U-Net is leveraged to post-processes the output image to generate some heatmaps (for keypoint detection) or masks (for segmentation)....

essunny310

Is there a printing error in the formula of the paper?

4

 Equation 1 here seems to be the instruction condition ci? Could the author confirm this?

houjie8888