Owen
Owen
I found it on [scripts/parse_landmark.py](https://github.com/TencentARC/GFPGAN/blob/master/scripts/parse_landmark.py) but the face_landmarks from `ffhq-dataset-v2.json`, so you need to get face_landmarks by dlib for your datasets
|权重|数据集|mAP| |-|-|-| |weights/yolov3.weights|coco|1.6121330694239914e-07| |checkpoints_coco_person/yolov3_ckpt_72.pth|coco_person|0.6085335843614279| |checkpoints_coco_person2/yolov3_ckpt_66.pth|coco_person|0.5898045606894886| |checkpoints/yolov3_ckpt_27.pth|caltech|0.21640950928418473| |checkpoints_caltech/yolov3_ckpt_29.pth|caltech|0.22963548969096598| I also train on coco for person class and caltech pedestrain detection ,after serveral epochs ,I test the weights, the result is very...
you can make it by dlib or face_alignment
> > > > > The official stylegan is trained on 4 batch * 8 GPUs. Therefore, I think it is better to use this setting to achieve the best...
you can use the code from [InstColorization/InstColorization.ipynb](https://github.com/ericsujw/InstColorization/blob/master/InstColorization.ipynb) ```pytrhon from google.colab.patches import cv2_imshow img_name_list = ['000000022969', '000000023781', '000000046872', '000000050145'] show_index = 1 img = cv2.imread('example/'+img_name_list[show_index]+'.jpg') lab_image = cv2.cvtColor(img, cv2.COLOR_BGR2LAB) l_channel, _,...
This is really great work. We have previously used a similar method employing CLIP for image search. Do you have a specific workflow? Thank you.
When using the pipe `from diffusers import StableDiffusionControlNetPipeline` and `from onediffx import compile_pipe`, or `from onediff.infer_compiler import oneflow_compile` in my code, I noticed a significant increase in GPU memory usage...
It is still useful. Previously, when I used `torch.cuda.empty_cache()`, it didn't work, but when I used `oneflow.cuda.empty_cache()`, it worked. However, the peak GPU memory usage during runtime remains the same....
> Thanks @onefish51 > > How many steps are you using? Seems that deepcache is not woking well with little steps( less than 30) as the workflow shown "20 steps"...