Added continuous labels

natanielruiz

2017-09-19 ec44ac453f794a5368e702315addfedcea3a4299

 code/batch_testing_preangles.py

@@ -61,7 +61,7 @@
    print 'Loading data.'

    transformations = transforms.Compose([transforms.Scale(224),
    transforms.RandomCrop(224), transforms.ToTensor(),
    transforms.CenterCrop(224), transforms.ToTensor(),
    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])])

    if args.dataset == 'AFLW2000':
@@ -114,12 +114,12 @@

        l1loss = torch.nn.L1Loss(size_average=False)

        for i, (images, labels, name) in enumerate(test_loader):
        for i, (images, labels, cont_labels, name) in enumerate(test_loader):
            images = Variable(images).cuda(gpu)
            total += labels.size(0)
            label_yaw = labels[:,0].float()
            label_pitch = labels[:,1].float()
            label_roll = labels[:,2].float()
            total += cont_labels.size(0)
            label_yaw = cont_labels[:,0].float()
            label_pitch = cont_labels[:,1].float()
            label_roll = cont_labels[:,2].float()

            yaw, pitch, roll, angles = model(images)

@@ -138,9 +138,9 @@
            roll_predicted = torch.sum(roll_predicted * idx_tensor, 1).cpu()

            # Mean absolute error
            yaw_error += torch.sum(torch.abs(yaw_predicted - label_yaw) * 3)
            pitch_error += torch.sum(torch.abs(pitch_predicted - label_pitch) * 3)
            roll_error += torch.sum(torch.abs(roll_predicted - label_roll) * 3)
            yaw_error += torch.sum(torch.abs(yaw_predicted * 3 - 99 - label_yaw))
            pitch_error += torch.sum(torch.abs(pitch_predicted * 3 - 99 - label_pitch))
            roll_error += torch.sum(torch.abs(roll_predicted * 3 - 99 - label_roll))

            if args.save_viz:
                name = name[0]