models_convit.py

# Copyright (c) 2022 Alpha-VL
# References:
# timm: https://github.com/rwightman/pytorch-image-models/tree/master/timm
# DeiT: https://github.com/facebookresearch/deit
# MAE: https://github.com/facebookresearch/mae
# ConvMAE: https://github.com/Alpha-VL/ConvMAE
# --------------------------------------------------------

from functools import partial

import torch
import torch.nn as nn
import pdb
import vision_transformer


class ConvViT(vision_transformer.ConvViT):
    """ Vision Transformer with support for global average pooling
    """
    def __init__(self, global_pool=False, **kwargs):
        super(ConvViT, self).__init__(**kwargs)
        self.global_pool = global_pool
        if self.global_pool:
            norm_layer = kwargs['norm_layer']
            embed_dim = kwargs['embed_dim']
            self.fc_norm = norm_layer(embed_dim[-1])

            del self.norm  # remove the original norm

    def forward_features(self, x):
        B = x.shape[0]
        x = self.patch_embed1(x)
        x = self.pos_drop(x)
        for blk in self.blocks1:
            x = blk(x)
        x = self.patch_embed2(x)
        for blk in self.blocks2:
            x = blk(x)
        x = self.patch_embed3(x)
        x = x.flatten(2).permute(0, 2, 1)
        x = self.patch_embed4(x)
        x = x + self.pos_embed
        for blk in self.blocks3:
            x = blk(x)
        if self.global_pool:
            x = x[:, :, :].mean(dim=1)  # global pool without cls token
            outcome = self.fc_norm(x)
        else:
            x = self.norm(x)
            outcome = x[:, 0]
        return outcome


def convit_base_patch16(**kwargs):
    model = ConvViT(
        img_size=[224, 56, 28], patch_size=[4, 2, 2], embed_dim=[256, 384, 768], depth=[2, 2, 11], num_heads=12, mlp_ratio=[8, 8, 4], qkv_bias=True,
        norm_layer=partial(nn.LayerNorm, eps=1e-6), **kwargs)
    return model


def convit_large_patch16(**kwargs):
    model = ConvViT(
        img_size=[224, 56, 28], patch_size=[4, 2, 2], embed_dim=[384, 768, 1024], depth=[2, 2, 23], num_heads=16, mlp_ratio=[8, 8, 4], qkv_bias=True,
        norm_layer=partial(nn.LayerNorm, eps=1e-6), **kwargs)
    return model