Function validate_sintel

evaluate_FlowFormer_tile.py:250–301

Peform validation using the Sintel (train) split

(model, sigma=0.05)

@torch.no_grad()
def validate_sintel(model, sigma=0.05):
    """ Peform validation using the Sintel (train) split """

    IMAGE_SIZE = [436, 1024]

    hws = compute_grid_indices(IMAGE_SIZE)
    weights = compute_weight(hws, IMAGE_SIZE, TRAIN_SIZE, sigma)

    model.eval()
    results = {}
    for dstype in ['final', "clean"]:
        val_dataset = datasets.MpiSintel(split='training', dstype=dstype)

        epe_list = []

        for val_id in range(len(val_dataset)):
            if val_id % 50 == 0:
                print(val_id)

            image1, image2, flow_gt, _ = val_dataset[val_id]
            image1 = image1[None].cuda()
            image2 = image2[None].cuda()

            flows = 0
            flow_count = 0

            for idx, (h, w) in enumerate(hws):
                image1_tile = image1[:, :, h:h+TRAIN_SIZE[0], w:w+TRAIN_SIZE[1]]
                image2_tile = image2[:, :, h:h+TRAIN_SIZE[0], w:w+TRAIN_SIZE[1]]

                flow_pre, _ = model(image1_tile, image2_tile, flow_init=None)

                padding = (w, IMAGE_SIZE[1]-w-TRAIN_SIZE[1], h, IMAGE_SIZE[0]-h-TRAIN_SIZE[0], 0, 0)
                flows += F.pad(flow_pre * weights[idx], padding)
                flow_count += F.pad(weights[idx], padding)

            flow_pre = flows / flow_count
            flow_pre = flow_pre[0].cpu()

            epe = torch.sum((flow_pre - flow_gt)**2, dim=0).sqrt()
            epe_list.append(epe.view(-1).numpy())

        epe_all = np.concatenate(epe_list)
        epe = np.mean(epe_all)
        px1 = np.mean(epe_all<1)
        px3 = np.mean(epe_all<3)
        px5 = np.mean(epe_all<5)

        print("Validation (%s) EPE: %f, 1px: %f, 3px: %f, 5px: %f" % (dstype, epe, px1, px3, px5))
        results[f"{dstype}_tile"] = np.mean(epe_list)

    return results

if __name__ == '__main__':
    parser = argparse.ArgumentParser()