resnet50.bamboo

A ResNet-50 pretrained on Bamboo-CLS, converted from the authors' release to native timm weights. No fine-tuning or weight rescaling was performed. The checkpoint wrapper and any module. prefix were removed. The released checkpoint has no classifier. num_classes=0 returns 2,048-dimensional pooled features.

Model details

• Paper: Bamboo
• Parameters: 23,508,032
• Input: RGB, 224 x 224
• Classifier outputs: 0
• Feature dimension: 2,048
• Training data: Bamboo-CLS

Preprocessing

Uses the authors' benchmark configuration: bicubic resize of the shorter side to 224, then a 224 x 224 center crop (crop_pct=1.0).

• Mean: (0.485, 0.456, 0.406)
• Standard deviation: (0.229, 0.224, 0.225)

The ViT benchmark uses CLIP normalization; the ResNet benchmark uses ImageNet normalization. The authors' separate ViT classification demo uses a different recipe. This export follows the benchmark.

Usage

import timm
import torch
from PIL import Image

model = timm.create_model("hf-hub:timm/resnet50.bamboo", pretrained=True).eval()
transform = timm.data.create_transform(**timm.data.resolve_model_data_config(model), is_training=False)
image = Image.open("image.jpg").convert("RGB")
with torch.inference_mode():
    output = model(transform(image).unsqueeze(0))

print(output.shape)  # torch.Size([1, 2048])

Source and licensing

Source URLs and SHA256 hashes are recorded in conversion.json. The authors' demo metadata declares CC-BY-4.0; the dataset access instructions state academic, non-commercial use. These are upstream statements, not a new license grant for the converted weights.

Citation

@misc{zhang2022bamboo,
  title={Bamboo: Building Mega-Scale Vision Dataset Continually with Human-Machine Synergy},
  author={Yuanhan Zhang and Qinghong Sun and Yichun Zhou and Zexin He and Zhenfei Yin and
           Kun Wang and Lu Sheng and Yu Qiao and Jing Shao and Ziwei Liu},
  year={2022},
  eprint={2203.07845},
  archivePrefix={arXiv},
  primaryClass={cs.CV}
}