models_vit.py

from functools import partial

import torch
import torch.nn as nn

import timm.models.vision_transformer

from util import recovery


class VisionTransformer(timm.models.vision_transformer.VisionTransformer):
    """ Vision Transformer with support for global average pooling
    """
    def __init__(self, global_pool=False, **kwargs):
        super(VisionTransformer, 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)

            del self.norm  # remove the original norm

    def forward_features(self, x, mask_idx=None):
        B = x.shape[0]

        # for certified recovery
        patch_mask = None

        if x.shape[1] == 4:
            patch_mask = recovery.mask_idx(x[:,3])
            x = x[:, :3]

        x = self.patch_embed(x)

        cls_tokens = self.cls_token.expand(B, -1, -1)  # stole cls_tokens impl from Phil Wang, thanks
        x = torch.cat((cls_tokens, x), dim=1)
        x = x + self.pos_embed

        if mask_idx != None:
            x = x[torch.tensor(list(range(B))).reshape(B,1).long(), mask_idx]

        if patch_mask is not None:
            patch_mask = patch_mask.unsqueeze(-1).expand(-1, -1, x.shape[2])
            x = torch.gather(x, 1, patch_mask)

        x = self.pos_drop(x)

        for blk in self.blocks:
            x = blk(x)

        if self.global_pool:
            x = x[:, 1:, :].mean(dim=1)  # global pool without cls token
            outcome = self.fc_norm(x)
        else:
            x = self.norm(x)
            outcome = x[:, 0]
        return outcome

    def forward(self, x, mask_idx=None):
        x = self.forward_features(x, mask_idx)
        x = self.head(x)
        return x

def vit_small_patch16(**kwargs):
    model = VisionTransformer(
        patch_size=16, embed_dim=384, depth=12, num_heads=6, mlp_ratio=4, qkv_bias=True,
        norm_layer=partial(nn.LayerNorm, eps=1e-6), **kwargs)
    return model


def vit_base_patch16(**kwargs):
    model = VisionTransformer(
        patch_size=16, embed_dim=768, depth=12, num_heads=12, mlp_ratio=4, qkv_bias=True,
        norm_layer=partial(nn.LayerNorm, eps=1e-6), **kwargs)
    return model


def vit_large_patch16(**kwargs):
    model = VisionTransformer(
        patch_size=16, embed_dim=1024, depth=24, num_heads=16, mlp_ratio=4, qkv_bias=True,
        norm_layer=partial(nn.LayerNorm, eps=1e-6), **kwargs)
    return model


def vit_huge_patch14(**kwargs):
    model = VisionTransformer(
        patch_size=14, embed_dim=1280, depth=32, num_heads=16, mlp_ratio=4, qkv_bias=True,
        norm_layer=partial(nn.LayerNorm, eps=1e-6), **kwargs)
    return model