Deepfake攻防 CV赛道Baseline理解 # Datawhale AI 夏令营

本文主要是利用chatGPT等工具，对baseline的内容代码进行理解和分析。

1.1 数据处理前的分析工作

'''
代码解释：统计文件的行数
！表示在jupyter中执行shell命令的符号
wc 表示一种统计文件的字数，行数的工具
-l 表示选项，统计文件的行数

'''

---

!wc -l /kaggle/input/deepfake/phase1/trainset_label.txt
!wc -l /kaggle/input/deepfake/phase1/valset_label.txt
!ls /kaggle/input/deepfake/phase1/trainset/ | wc -l 
!ls /kaggle/input/deepfake/phase1/valset/ | wc -l 

1.2 数据路径的处理

# 读取到每一个图像的路径

train_label = pd.read_csv('/kaggle/input/deepfake/phase1/trainset_label.txt')
val_label = pd.read_csv('/kaggle/input/deepfake/phase1/valset_label.txt')

train_label['path'] = '/kaggle/input/deepfake/phase1/trainset/' + train_label['img_name']
val_label['path'] = '/kaggle/input/deepfake/phase1/valset/' + val_label['img_name']

1.3 对于数据集的处理工作

#  validate函数，用于在验证集上评估深度学习模型的性能

#  model.eval() 将模型切换到评估模式

# 遍历验证数据加载器

# 训练集和验证集的代码相似

---

def validate(val_loader, model, criterion):
    batch_time = AverageMeter('Time', ':6.3f')
    losses = AverageMeter('Loss', ':.4e')
    top1 = AverageMeter('Acc@1', ':6.2f')
    progress = ProgressMeter(len(val_loader), batch_time, losses, top1)

    # switch to evaluate mode
    model.eval()

    with torch.no_grad():
        end = time.time()
        for i, (input, target) in tqdm_notebook(enumerate(val_loader), total=len(val_loader)):
            input = input.cuda()
            target = target.cuda()

            # compute output
            output = model(input)
            loss = criterion(output, target)

            # measure accuracy and record loss
            acc = (output.argmax(1).view(-1) == target.float().view(-1)).float().mean() * 100
            losses.update(loss.item(), input.size(0))
            top1.update(acc, input.size(0))
            # measure elapsed time
            batch_time.update(time.time() - end)
            end = time.time()

        # TODO: this should also be done with the ProgressMeter
        print(' * Acc@1 {top1.avg:.3f}'
              .format(top1=top1))
        return top1