MixMIM: Mixed and Masked Image Modeling for Efficient Visual Representation Learning
In this study, we propose Mixed and Masked Image Modeling (MixMIM), a simple but efficient MIM method that is applicable to various hierarchical Vision Transformers. Existing MIM methods replace a random subset of input tokens with a special [MASK] symbol and aim at reconstructing original image tokens from the corrupted image. However, we find that using the [MASK] symbol greatly slows down the training and causes training-finetuning inconsistency, due to the large masking ratio (e.g., 40% in BEiT). In contrast, we replace the masked tokens of one image with visible tokens of another image, i.e., creating a mixed image. We then conduct dual reconstruction to reconstruct the original two images from the mixed input, which significantly improves efficiency. While MixMIM can be applied to various architectures, this paper explores a simpler but stronger hierarchical Transformer, and scales with MixMIM-B, -L, and -H. Empirical results demonstrate that MixMIM can learn high-quality visual representations efficiently. Notably, MixMIM-B with 88M parameters achieves 85.1% top-1 accuracy on ImageNet-1K by pretraining for 600 epochs, setting a new record for neural networks with comparable model sizes (e.g., ViT-B) among MIM methods. Besides, its transferring performances on the other 6 datasets show MixMIM has better FLOPs / performance tradeoff than previous MIM methods
Predict image
from mmpretrain import inference_model
predict = inference_model('mixmim-base_mixmim-pre_8xb128-coslr-100e_in1k', 'demo/bird.JPEG')
print(predict['pred_class'])
print(predict['pred_score'])
Use the model
import torch
from mmpretrain import get_model
model = get_model('mixmim_mixmim-base_16xb128-coslr-300e_in1k', pretrained=True)
inputs = torch.rand(1, 3, 224, 224)
out = model(inputs)
print(type(out))
# To extract features.
feats = model.extract_feat(inputs)
print(type(feats))
Train/Test Command
Prepare your dataset according to the docs.
Train:
python tools/train.py configs/mixmim/mixmim_mixmim-base_16xb128-coslr-300e_in1k.py
Test:
python tools/test.py configs/mixmim/benchmarks/mixmim-base_8xb128-coslr-100e_in1k.py https://download.openmmlab.com/mmselfsup/1.x/mixmim/mixmim-base-p16_16xb128-coslr-300e_in1k/mixmim-base-p16_ft-8xb128-coslr-100e_in1k/mixmim-base-p16_ft-8xb128-coslr-100e_in1k_20221208-41ecada9.pth
Model | Params (M) | Flops (G) | Config | Download |
---|---|---|---|---|
mixmim_mixmim-base_16xb128-coslr-300e_in1k |
114.67 | 16.35 | config | model | log |
Model | Pretrain | Params (M) | Flops (G) | Top-1 (%) | Config | Download |
---|---|---|---|---|---|---|
mixmim-base_mixmim-pre_8xb128-coslr-100e_in1k |
MIXMIM | 88.34 | 16.35 | 84.63 | config | model | log |
@article{MixMIM2022,
author = {Jihao Liu, Xin Huang, Yu Liu, Hongsheng Li},
journal = {arXiv:2205.13137},
title = {MixMIM: Mixed and Masked Image Modeling for Efficient Visual Representation Learning},
year = {2022},
}