garychan22
garychan22
the provided demo in mmpose4aigc only outputs 18 keypoints while the https://github.com/CMU-Perceptual-Computing-Lab outputs 25 keypoints, which model should I use to obtain 25 keypoints ? thx
@Z-YuPeng Hi，你好，我想问一下关于random sampling为什么会work，因为其他相关工作一般会通过cross-attention来得到物体mask然后refer到这些特定区域的特征，还是说更dropout的意味差不多，谢谢
@bssrdf 兄弟，这一点不像杨幂hhhh
installing mmcv following the cmd does not work either `pip install mmcv==2.2.0 -f https://download.openmmlab.com/mmcv/dist/cu121/torch2.3/index.html` Instead, I install the mmsegmentation locally and change `MMCV_MAX = '2.2.0'` to `MMCV_MAX = '2.2.1'`, and...
@YUHANG-Ma Hi, any plan to release the refined training data?
@ToTheBeginning did you use the flux-schnell model as the accelerated model?
still getting blurred faces. I am wondering do we need to align the face as in FFHQ before seeding our own images to the inference script
I am training pulid like non-acclerated model for diffusion loss and acclerated model for id loss, and a weird problem arises: the generated image includes many faces of the same...
I am using the following script to process about 10M images ``` import argparse import json import os import shutil import uuid from tqdm import tqdm import glob2 import torch...