Trying superres

natanielruiz

2017-09-23 31fc66b795c0a57b8009d7b03f49f6cd099ceb29

 code/test_preangles.py

@@ -67,10 +67,15 @@
    if args.dataset == 'AFLW2000':
        pose_dataset = datasets.AFLW2000(args.data_dir, args.filename_list,
                                transformations)
    elif args.dataset == 'AFLW2000_ds':
        pose_dataset = datasets.AFLW2000_ds(args.data_dir, args.filename_list,
                                transformations)
    elif args.dataset == 'BIWI':
        pose_dataset = datasets.BIWI(args.data_dir, args.filename_list, transformations)
    elif args.dataset == 'AFLW':
        pose_dataset = datasets.AFLW(args.data_dir, args.filename_list, transformations)
    elif args.dataset == 'Pose_300W_LP':
        pose_dataset = datasets.Pose_300W_LP(args.data_dir, args.filename_list, transformations)
    elif args.dataset == 'AFW':
        pose_dataset = datasets.AFW(args.data_dir, args.filename_list, transformations)
    else:
@@ -97,12 +102,12 @@

    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)

@@ -116,21 +121,27 @@
        pitch_predicted = utils.softmax_temperature(pitch.data, 1)
        roll_predicted = utils.softmax_temperature(roll.data, 1)

        yaw_predicted = torch.sum(yaw_predicted * idx_tensor, 1).cpu()
        pitch_predicted = torch.sum(pitch_predicted * idx_tensor, 1).cpu()
        roll_predicted = torch.sum(roll_predicted * idx_tensor, 1).cpu()
        yaw_predicted = torch.sum(yaw_predicted * idx_tensor, 1).cpu() * 3 - 99
        pitch_predicted = torch.sum(pitch_predicted * idx_tensor, 1).cpu() * 3 - 99
        roll_predicted = torch.sum(roll_predicted * idx_tensor, 1).cpu() * 3 - 99

        # 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 - label_yaw))
        pitch_error += torch.sum(torch.abs(pitch_predicted - label_pitch))
        roll_error += torch.sum(torch.abs(roll_predicted - label_roll))

        # Save images with pose cube.
        # TODO: fix for larger batch size
        if args.save_viz:
            name = name[0]
            cv2_img = cv2.imread(os.path.join(args.data_dir, name + '.jpg'))
            utils.plot_pose_cube(cv2_img, yaw_predicted[0] * 3 - 99, pitch_predicted[0] * 3 - 99, roll_predicted[0] * 3 - 99)
            if args.dataset == 'BIWI':
                cv2_img = cv2.imread(os.path.join(args.data_dir, name + '_rgb.png'))
            else:
                cv2_img = cv2.imread(os.path.join(args.data_dir, name + '.jpg'))
            if args.batch_size == 1:
                error_string = 'y %.2f, p %.2f, r %.2f' % (torch.sum(torch.abs(yaw_predicted - label_yaw)), torch.sum(torch.abs(pitch_predicted - label_pitch)), torch.sum(torch.abs(roll_predicted - label_roll)))
                cv2.putText(cv2_img, error_string, (30, cv2_img.shape[0]- 30), fontFace=1, fontScale=1, color=(0,0,255), thickness=1)
            utils.plot_pose_cube(cv2_img, yaw_predicted[0], pitch_predicted[0], roll_predicted[0])
            cv2.imwrite(os.path.join('output/images', name + '.jpg'), cv2_img)

    print('Test error in degrees of the model on the ' + str(total) +