IGEV深度估计测试代码

news2024/11/25 10:24:43

生成点云的测试脚本：

import sys
sys.path.append('core')
DEVICE = 'cuda'
import os

os.environ['CUDA_VISIBLE_DEVICES'] = '0'
import argparse
import glob
import numpy as np
import torch
from tqdm import tqdm
from pathlib import Path
from igev_stereo import IGEVStereo
from utils.utils import InputPadder
from PIL import Image
from matplotlib import pyplot as plt
import os
import cv2
import time

def load_image(imfile):
    img = np.array(Image.open(imfile)).astype(np.uint8)
    img = torch.from_numpy(img).permute(2, 0, 1).float()
    return img[None].to(DEVICE)
def load_image1(img):
    # img = np.array(Image.open(imfile)).astype(np.uint8)
    img = torch.from_numpy(img).permute(2, 0, 1).float()
    return img[None].to(DEVICE)

def disp2pointcloud(disp, K, disline):
    width = disp.shape[1]
    height = disp.shape[0]
    f = open('pointcloud.txt','w')
    print(K[0, 2])
    for i in range(width):
        for j in range(height):
            z = K[0, 0]/disp[j, i]*disline
            x = (i-K[0, 2])/K[0, 0]*z
            y = (j-K[1, 2])/K[1, 1]*z
            f.write('%f %f %f\n' %(x, y, z))
    f.close()


def demo(args):
    model = torch.nn.DataParallel(IGEVStereo(args), device_ids=[0])
    model.load_state_dict(torch.load(args.restore_ckpt))

    model = model.module
    model.to(DEVICE)
    model.eval()

    output_directory = Path(args.output_directory)
    output_directory.mkdir(exist_ok=True)

    with torch.no_grad():
        left_images = sorted(glob.glob(args.left_imgs, recursive=True))
        right_images = sorted(glob.glob(args.right_imgs, recursive=True))
        print(f"Found {len(left_images)} images. Saving files to {output_directory}/")
        
        for (imfile1, imfile2) in tqdm(list(zip(left_images, right_images))):
            # image1 = load_image(imfile1)
            # image2 = load_image(imfile2)
            # print(imfile2)
            start = time.time()
            image1 = cv2.imread(imfile1)
            image2 = cv2.imread(imfile2)

            width = int(image1.shape[1] * 0.5)
            height = int(image1.shape[0] * 0.5)

            dim = (width, height)
            image1 = cv2.resize(image1, dim, interpolation = cv2.INTER_AREA)
            image2 = cv2.resize(image2, dim, interpolation = cv2.INTER_AREA)
            #np.asarray(bytearray(req.read()), dtype=np.uint8)
            # image1 = np.asarray(image1)
            # image2 = np.asarray(image2)
            image1 = load_image1(image1)
            image2 = load_image1(image2)

            padder = InputPadder(image1.shape, divis_by=32)
            image1, image2 = padder.pad(image1, image2)

            disp = model(image1, image2, iters=args.valid_iters, test_mode=True)
            disp = disp.cpu().numpy()
            
            K = np.array([[2.4219981e+03*0.5, 0, 1.2478e+3*0.5], [0, 2.4219981e+03*0.5, 1.00927e+3*0.5],[0, 0, 1]])
            disline = 1.4939830129441067e+00
            disp2pointcloud(disp.squeeze(), K, disline)
            disp = padder.unpad(disp)
            end = time.time()
            print(end-start)
            file_stem = imfile1.split('/')[-2]
            filename = os.path.join(output_directory, f"{file_stem}.png")
            plt.imsave(output_directory / f"{file_stem}.png", disp.squeeze(), cmap='jet')
            
            # disp = np.round(disp * 256).astype(np.uint16)
            # cv2.imwrite(filename, cv2.applyColorMap(cv2.convertScaleAbs(disp.squeeze(), alpha=0.01),cv2.COLORMAP_JET), [int(cv2.IMWRITE_PNG_COMPRESSION), 0])


if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('--restore_ckpt', help="restore checkpoint", default='./pretrained_models/sceneflow/sceneflow.pth')
    parser.add_argument('--save_numpy', action='store_true', help='save output as numpy arrays')

    parser.add_argument('-l', '--left_imgs', help="path to all first (left) frames", default="./demo-imgs/leador/0.jpg")
    parser.add_argument('-r', '--right_imgs', help="path to all second (right) frames", default="./demo-imgs/leador/1.jpg")

    # parser.add_argument('-l', '--left_imgs', help="path to all first (left) frames", default="/data/Middlebury/trainingH/*/im0.png")
    # parser.add_argument('-r', '--right_imgs', help="path to all second (right) frames", default="/data/Middlebury/trainingH/*/im1.png")
    # parser.add_argument('-l', '--left_imgs', help="path to all first (left) frames", default="/data/ETH3D/two_view_training/*/im0.png")
    # parser.add_argument('-r', '--right_imgs', help="path to all second (right) frames", default="/data/ETH3D/two_view_training/*/im1.png")
    parser.add_argument('--output_directory', help="directory to save output", default="./demo-output/")
    parser.add_argument('--mixed_precision', action='store_true', help='use mixed precision')
    parser.add_argument('--valid_iters', type=int, default=32, help='number of flow-field updates during forward pass')

    # Architecture choices
    parser.add_argument('--hidden_dims', nargs='+', type=int, default=[128]*3, help="hidden state and context dimensions")
    parser.add_argument('--corr_implementation', choices=["reg", "alt", "reg_cuda", "alt_cuda"], default="reg", help="correlation volume implementation")
    parser.add_argument('--shared_backbone', action='store_true', help="use a single backbone for the context and feature encoders")
    parser.add_argument('--corr_levels', type=int, default=2, help="number of levels in the correlation pyramid")
    parser.add_argument('--corr_radius', type=int, default=4, help="width of the correlation pyramid")
    parser.add_argument('--n_downsample', type=int, default=2, help="resolution of the disparity field (1/2^K)")
    parser.add_argument('--slow_fast_gru', action='store_true', help="iterate the low-res GRUs more frequently")
    parser.add_argument('--n_gru_layers', type=int, default=3, help="number of hidden GRU levels")
    parser.add_argument('--max_disp', type=int, default=192, help="max disp of geometry encoding volume")
    
    args = parser.parse_args()
    Path(args.output_directory).mkdir(exist_ok=True, parents=True)

    demo(args)