- Merge 2 open-source datasets:
flower_dataset.py - Generate ImageNet-like dataset for competition data:
imnetfy_ds.py
How to load pretrain VOLO model:
- Clone actual model:
git clone https://github.com/sail-sg/volo - Download weights: link
conda install timmorpip install timm- Load the model:
import torch
from volo.models.volo import *
checkpoint_path = "weights_volo/volo_d1_244_competition_original_dataset/model_best.pth.tar"
# create model
model = volo_d1(img_size=224, num_classes=219)
# load weights
checkpoint = torch.load(checkpoint_path, map_location='cpu')
state_dict = checkpoint['state_dict']
model.load_state_dict(state_dict)
# display the model
print(model)
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent