File size: 6,607 Bytes
2df812d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
'''
Author: Chris Xiao yl.xiao@mail.utoronto.ca
Date: 2023-09-16 19:47:31
LastEditors: Chris Xiao yl.xiao@mail.utoronto.ca
LastEditTime: 2023-09-30 16:56:23
FilePath: /EndoSAM/endoSAM/model.py
Description: EndoSAM model adapter 
I Love IU
Copyright (c) 2023 by Chris Xiao yl.xiao@mail.utoronto.ca, All Rights Reserved. 
'''
import torch
import torch.nn as nn
from einops import rearrange
from utils import postprocess_masks

class EndoSAMAdapter(nn.Module):
    def __init__(self, device, 
                 num_classes,
                 sam_mask_encoder,
                 sam_prompt_encoder,
                 sam_mask_decoder,
                 num_token=8,
                 ):
        super(EndoSAMAdapter, self).__init__()
        self.device = device
        self.num_classes = num_classes - 1
        self.num_token = num_token
        self.sam_mask_encoder = sam_mask_encoder.to(self.device)
        self.sam_prompt_encoder = sam_prompt_encoder.to(self.device)
        self.sam_mask_decoder = sam_mask_decoder.to(self.device)
        self.prototype_prompt_encoder = Prototype_Prompt_Encoder(feat_dim=256, 
                                                            hidden_dim_dense=128, 
                                                            hidden_dim_sparse=128, 
                                                            size=64, 
                                                            num_tokens=self.num_token).to(self.device)
        self.learnable_prototypes_model = Learnable_Prototypes(num_classes=self.num_classes, feat_dim = 256).to(self.device)
        self.prototypes = self.learnable_prototypes_model()
        self.sam_mask_encoder.to(self.device)
        self.sam_prompt_encoder.to(self.device)
        self.sam_mask_decoder.to(self.device)
        
        for _, param in self.prototype_prompt_encoder.named_parameters():
            param.requires_grad = True
        
        for _, param in self.learnable_prototypes_model.named_parameters():
            param.requires_grad = True
        
        for _, param in self.sam_mask_decoder.named_parameters():
            param.requires_grad = True
        
        for _, param in self.sam_mask_encoder.named_parameters():
            param.requires_grad = False
        
        for _, param in self.sam_prompt_encoder.named_parameters():
            param.requires_grad = False


    def forward(self, x):
        sam_features = self.sam_mask_encoder(x)
        sam_features = rearrange(sam_features, 'b c h w -> b (h w) c')
        cls_ids = torch.tensor(1).repeat(sam_features.shape[0]).to(self.device)
        dense_embeddings, sparse_embeddings = self.prototype_prompt_encoder(sam_features, self.prototypes, cls_ids, self.num_classes)
        pred = []
        pred_quality = []
        sam_features = rearrange(sam_features,'b (h w) c -> b c h w', h=64, w=64)
        for dense_embedding, sparse_embedding, features_per_image in zip(dense_embeddings.unsqueeze(1), sparse_embeddings.unsqueeze(1), sam_features):    
            low_res_masks_per_image, mask_quality_per_image = self.sam_mask_decoder(
                    image_embeddings=features_per_image.unsqueeze(0),
                    image_pe=self.sam_prompt_encoder.get_dense_pe(), 
                    sparse_prompt_embeddings=sparse_embedding,
                    dense_prompt_embeddings=dense_embedding, 
                    multimask_output=True,
                )
            pred_per_image = postprocess_masks(
                low_res_masks_per_image,
                input_size=(819, 1024),
                original_size=(1024, 1280),
            )
            
            pred.append(pred_per_image)
            pred_quality.append(mask_quality_per_image)
        
        pred = torch.cat(pred, dim=0)
        pred_quality = torch.cat(pred_quality,dim=0)
        return pred, pred_quality

class Prototype_Prompt_Encoder(nn.Module):
    def __init__(self, feat_dim=256, 
                 hidden_dim_dense=128, 
                 hidden_dim_sparse=128, 
                 size=64, 
                 num_tokens=8):
                
        super(Prototype_Prompt_Encoder, self).__init__()
        self.dense_fc_1 = nn.Conv2d(feat_dim, hidden_dim_dense, 1)
        self.dense_fc_2 = nn.Conv2d(hidden_dim_dense, feat_dim, 1)
        
        self.relu = nn.ReLU()

        self.sparse_fc_1 = nn.Conv1d(size*size, hidden_dim_sparse, 1)
        self.sparse_fc_2 = nn.Conv1d(hidden_dim_sparse, num_tokens, 1)
        
        
        pn_cls_embeddings = [nn.Embedding(num_tokens, feat_dim) for _ in range(2)] # one for positive and one for negative 

            
        self.pn_cls_embeddings = nn.ModuleList(pn_cls_embeddings)
                
    def forward(self, feat, prototypes, cls_ids, num_classes):
  
        cls_prompts = prototypes.unsqueeze(-1)
        cls_prompts = torch.stack([cls_prompts for _ in range(feat.size(0))], dim=0)

        
        feat = torch.stack([feat for _ in range(cls_prompts.size(1))], dim=1)
        # compute similarity matrix 
        sim = torch.matmul(feat, cls_prompts)
        
        # compute class-activated feature
        feat =  feat + feat*sim
        feat_sparse = feat.clone()
        
        # compute dense embeddings
        one_hot = torch.nn.functional.one_hot(cls_ids-1,num_classes)
        feat = feat[one_hot == 1]
        feat = rearrange(feat.squeeze(1),'b (h w) c -> b c h w', h=64, w=64)
        dense_embeddings = self.dense_fc_2(self.relu(self.dense_fc_1(feat)))
        
        # compute sparse embeddings
        feat_sparse = rearrange(feat_sparse,'b num_cls hw c -> (b num_cls) hw c')
        sparse_embeddings = self.sparse_fc_2(self.relu(self.sparse_fc_1(feat_sparse)))
        sparse_embeddings = rearrange(sparse_embeddings,'(b num_cls) n c -> b num_cls n c', num_cls=1)
        
        pos_embed = self.pn_cls_embeddings[1].weight.unsqueeze(0).unsqueeze(0) * one_hot.unsqueeze(-1).unsqueeze(-1)
        neg_embed = self.pn_cls_embeddings[0].weight.unsqueeze(0).unsqueeze(0) * (1-one_hot).unsqueeze(-1).unsqueeze(-1)
        

        sparse_embeddings = sparse_embeddings + pos_embed.detach() + neg_embed.detach()
            
        sparse_embeddings = rearrange(sparse_embeddings,'b num_cls n c -> b (num_cls n) c')
        
        return dense_embeddings, sparse_embeddings 


class Learnable_Prototypes(nn.Module):
    def __init__(self, num_classes=7 , feat_dim=256):
        super(Learnable_Prototypes, self).__init__()
        self.class_embeddings = nn.Embedding(num_classes, feat_dim)
        
    def forward(self):
        return self.class_embeddings.weight