CSLR/seq_scripts.py at main · tjyyy3/CSLR · GitHub

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
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
import os
import pdb
import sys
import copy
import torch
import numpy as np
import torch.nn as nn
from tqdm import tqdm
import torch.nn.functional as F
import matplotlib.pyplot as plt
from evaluation.slr_eval.wer_calculation import evaluate
from evaluation.slt_eval.txt_calculation import bleu, chrf, rouge, wer_list

def seq_train(loader, model, optimizer, device, epoch_idx, recoder):  # 训练
    model.train()
    loss_value = []
    clr = [group['lr'] for group in optimizer.optimizer.param_groups]
    for batch_idx, data in enumerate(loader):  # 循环遍历数据集加载器(loader)中的每个批次数据
        vid = device.data_to_device(data[0])
        vid_lgt = device.data_to_device(data[1])
        label = device.data_to_device(data[2])
        label_lgt = device.data_to_device(data[3])
        ret_dict = model(vid, vid_lgt, label=label, label_lgt=label_lgt)
        loss = model.criterion_calculation(ret_dict, label, label_lgt)   # 损失
        if np.isinf(loss.item()) or np.isnan(loss.item()):
            print(data[-1])
            continue
        optimizer.zero_grad()   # 优化器梯度清零
        loss.backward()    # 对损失进行反向传播
        # nn.utils.clip_grad_norm_(model.rnn.parameters(), 5)
        optimizer.step()
        loss_value.append(loss.item())
        if batch_idx % recoder.log_interval == 0:
            recoder.print_log(
                '\tEpoch: {}, Batch({}/{}) done. Loss: {:.8f}  lr:{:.6f}'    # 每50个batch打印一个日志：损失
                    .format(epoch_idx, batch_idx, len(loader), loss.item(), clr[0]))
    optimizer.scheduler.step()    # 在一个epoch要结束时，调整学习率
    recoder.print_log('\tMean training loss: {:.10f}.'.format(np.mean(loss_value)))
    return loss_value


def seq_eval(cfg, loader, model, device, mode, epoch, work_dir, recoder, evaluate_tool="python"):  # 测试
    model.eval()
    total_sent = []   # 预测值
    total_info = []
    loss_value = []   # 每个batch的loss值
    total_conv_sent = []
    wer = []
    stat = {i: [0, 0] for i in range(len(loader.dataset.dict))}
    for batch_idx, data in enumerate(tqdm(loader)):
        recoder.record_timer("device")
        vid = device.data_to_device(data[0])  # 视频数据
        vid_lgt = device.data_to_device(data[1])  # 视频长度
        label = device.data_to_device(data[2])  # 标签数据
        label_lgt = device.data_to_device(data[3])  # 标签长度
        with torch.no_grad():
            ret_dict = model(vid, vid_lgt, label=label, label_lgt=label_lgt)   # 预测结果(词)
            loss = model.criterion_calculation(ret_dict, label, label_lgt)
            loss_value.append(loss.item())

        total_info += [file_name.split("/")[0] for file_name in data[-1]]   # 测试数据的标签

        all_gloss_label = data[-1]
        total_sent += ret_dict['recognized_sents']   # 预测结果（句子）
        total_conv_sent += ret_dict['conv_sents']
        total_merge_sent = ret_dict['merge_sents']  # 模型预测的句子
        gloss_hyp = []  # 模型预测的词汇序列

        for sample_idx, sample in enumerate(total_merge_sent):
            x = []
            for word_idx, word in enumerate(sample):
                x.append(word[0])
            gloss_hyp += [' '.join(x).upper()]
        gloss_ref = [''.join(t) for t in all_gloss_label]
        gloss_wer = wer_list(references=gloss_ref, hypotheses=gloss_hyp)
        wer.append(round(gloss_wer["wer"], 2))

    try:
        python_eval = True if evaluate_tool == "python" else False
        write2file(work_dir + "output-hypothesis-{}.ctm".format(mode), total_info, total_sent)
        write2file(work_dir + "output-hypothesis-{}-conv.ctm".format(mode), total_info, total_conv_sent)
        write2file(work_dir + "output-hypothesis-{}-merge.ctm".format(mode), total_info, total_merge_sent)

        conv_ret = evaluate(
            evaluate_dir=cfg.dataset_info['evaluation_dir'],
            evaluate_prefix=cfg.dataset_info['evaluation_prefix'],
            output_dir="epoch_{}_result/".format(epoch),
            python_evaluate=python_eval,
        )
        lstm_ret = evaluate(
            prefix=work_dir, mode=mode, output_file="result.txt",
            evaluate_dir=cfg.dataset_info['evaluation_dir'],
            evaluate_prefix=cfg.dataset_info['evaluation_prefix'],
            output_dir="epoch_{}_result/".format(epoch),
            python_evaluate=python_eval,
            triplet=True,
        )
        merge_ret = evaluate(
            prefix=work_dir, mode=mode, output_file="output-hypothesis-{}-merge.ctm".format(mode),
            evaluate_dir=cfg.dataset_info['evaluation_dir'],
            evaluate_prefix=cfg.dataset_info['evaluation_prefix'],
            output_dir="epoch_{}_result/".format(epoch),
            python_evaluate=python_eval,
        )

    except:
        print("Unexpected error:", sys.exc_info()[0])
        lstm_ret = 100.0
    finally:
        pass
        # recoder.print_log(f"Epoch {epoch}, {mode} {conv_ret: 2.2f}%", f"{work_dir}/{mode}.txt")
        # recoder.print_log(f"Epoch {epoch}, {mode} {lstm_ret: 2.2f}%", f"{work_dir}/{mode}.txt")
        # recoder.print_log(f"Epoch {epoch}, {mode} {merge_ret: 2.2f}%", f"{work_dir}/{mode}.txt")
        recoder.print_log(f"Epoch {epoch}, {mode} {np.mean(wer): 2.2f}%", f"{work_dir}/{mode}.txt")
        recoder.print_log('\tMean Dev loss: {:.10f}.'.format(np.mean(loss_value)))
        # return conv_ret, lstm_ret, merge_ret
        return wer


def seq_feature_generation(loader, model, device, mode, work_dir, recoder):
    model.eval()

    src_path = os.path.abspath(f"{work_dir}{mode}")
    tgt_path = os.path.abspath(f"./features/{mode}")
    if not os.path.exists("./features/"):
    	os.makedirs("./features/")

    if os.path.islink(tgt_path):
        curr_path = os.readlink(tgt_path)
        if work_dir[1:] in curr_path and os.path.isabs(curr_path):
            return
        else:
            os.unlink(tgt_path)
    else:
        if os.path.exists(src_path) and len(loader.dataset) == len(os.listdir(src_path)):
            os.symlink(src_path, tgt_path)
            return

    for batch_idx, data in tqdm(enumerate(loader)):
        recoder.record_timer("device")
        vid = device.data_to_device(data[0])
        vid_lgt = device.data_to_device(data[1])
        with torch.no_grad():
            ret_dict = model(vid, vid_lgt)
        if not os.path.exists(src_path):
            os.makedirs(src_path)
        start = 0
        for sample_idx in range(len(vid)):
            end = start + data[3][sample_idx]
            filename = f"{src_path}/{data[-1][sample_idx].split('|')[0]}_features.npy"
            save_file = {
                "label": data[2][start:end],
                "features": ret_dict['framewise_features'][sample_idx][:, :vid_lgt[sample_idx]].T.cpu().detach(),
            }
            np.save(filename, save_file)
            start = end
        assert end == len(data[2])
    os.symlink(src_path, tgt_path)


def write2file(path, info, output):
    filereader = open(path, "w")
    for sample_idx, sample in enumerate(output):
        for word_idx, word in enumerate(sample):
            filereader.writelines(
                "{} 1 {:.2f} {:.2f} {}\n".format(info[sample_idx],
                                                 word_idx * 1.0 / 100,
                                                 (word_idx + 1) * 1.0 / 100,
                                                 word[0]))