model.py

import torch
import torch.nn as nn
import torchvision.models as models
from torch.nn.utils.rnn import pack_padded_sequence


class EncoderCNN(nn.Module):
    def __init__(self, embed_size):
        super(EncoderCNN, self).__init__()
        resnet = models.resnet50(pretrained=True)
        for param in resnet.parameters():
            param.requires_grad_(False)
        
        modules = list(resnet.children())[:-1]
        self.resnet = nn.Sequential(*modules)
        self.embed = nn.Linear(resnet.fc.in_features, embed_size)

    def forward(self, images):
        features = self.resnet(images)
        features = features.view(features.size(0), -1)
        features = self.embed(features)
        return features
    

class DecoderRNN(nn.Module):
    def __init__(self, embed_size, hidden_size, vocab_size, dropout=0.3, num_layers=2):
        super(DecoderRNN, self).__init__()
        self.hidden_dim = hidden_size
        self.embed = nn.Embedding(vocab_size, embed_size)
        self.dropout = dropout
        self.lstm = nn.LSTM(embed_size, hidden_size, num_layers, batch_first=True, dropout=dropout)
        self.linear = nn.Linear(hidden_size, vocab_size)
        self.dropout_layer = nn.Dropout(p=dropout)
        self.hidden = (torch.zeros(1, 1, hidden_size),torch.zeros(1, 1, hidden_size))
        
    
    def forward(self, features, captions):
        """Decode image feature vectors and generates captions."""
        embeddings = self.embed(captions[:,:-1])
        embeddings = torch.cat((features.unsqueeze(1), embeddings), 1)
        print(embeddings.shape)
        lstm_out, self.hidden = self.lstm(embeddings)
        outputs = self.linear(lstm_out)
        print(outputs.shape)
        return outputs

    def sample(self, inputs,states=None, max_seg_len=20):
        " accepts pre-processed image tensor (inputs) and returns predicted sentence (list of tensor ids of length max_len) "
        sampled_ids = []
        
        for i in range(max_seg_len):
            hiddens, states = self.lstm(inputs, states)         
            outputs = self.linear(hiddens.squeeze(1))            
            _, predicted = outputs.max(1)
            sampled_ids.append(predicted.item())
            inputs = self.embed(predicted)
            inputs = inputs.unsqueeze(1)                         
                
        return sampled_ids