VIL100 dataset processing

VIL100 dataset (IMG:image(size ((1080 * 1920), (720 * 1280), (480 * 960), (478 * 960), (448 * 960), (378 * 672), (474 ​​* 960 )), GT:json)

Purpose: to process into a uniform size, which is convenient for model train.
(resize to 800 * 320 after clipping based on the vanishing point of the lane line, GT format includes mask, line_txt in point format)

import os
import glob
import shutil
import cv2
import json
import numpy as np
path="/VIL100"
file_path="/VIL100/data/train.txt"
img_path="/merge/VIL100/image"
mask_path="/merge/VIL100/mask"
text_path = '/merge/VIL100/line_txt/'
n_0 =0
new_w = 800
new_h = 320
with open(file_path,'r') as f:
    for line in f:
        image_name_old = line.split()[0].split('/')[-1]
        ori_image = cv2.imread(path+line.split()[0])
        mask_name_old = line.split()[1].split('/')[-1]
        # read json
        json_name = path + line.split()[0].replace('JPEGImages', 'Json')+'.json'
        mask_json = json.load((open(json_name, encoding='utf-8')))
        anno = mask_json['annotations']['lane']
        points_list = []
        for anno_line in anno:
            points_list.append(np.array(anno_line['points']))
        if len(anno) == 0:
            print("@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@kong")
            print(points_list)
            continue
        mask =  cv2.imread(path+line.split()[1],0)
        Num= np.amax(mask[:,:],axis=0)
        dice=np.argmax(mask[:,:],axis=0)
        used_dice=[]
        for i in range(0,len(dice)):
            if dice[i]!=0:
                used_dice.append(dice[i])
        if len(used_dice)==0:
            used_dice.append(0)
        crop_y=np.amin(used_dice)
        # print(crop_y)
        img = np.zeros((mask.shape[0]-crop_y, mask.shape[1], 3), np.uint8)
        img[:mask.shape[0]-crop_y, :, :] = ori_image[crop_y:, ...]
        mas = np.zeros((mask.shape[0] - crop_y, mask.shape[1], 3), np.uint8)
        mas[:mask.shape[0] - crop_y, :, 0] = mask[crop_y:, :]
        mas[:mask.shape[0] - crop_y, :, 1] = mask[crop_y:, :]
        mas[:mask.shape[0] - crop_y, :, 2] = mask[crop_y:, :]
        img_new = cv2.resize(img, (new_w, new_h), interpolation=1)
        mas_new = cv2.resize(mas, (new_w, new_h), interpolation=1)
        draw_img = img_new.copy()
        mas_new =np.where(mas_new>0 ,255,0)
        len_num = len(str(n_0))
        split_0 = 8 - len_num
        Public_name = "0000000000000000000000000000000000000000000000000000000000000000"
        split_name = Public_name[0:split_0]
        #Visualization crop and resize for the points of json
        # for idx, pts in enumerate(points_list):
        #     for i in range(1, len(pts)):
        #         x1 = int(pts[i - 1][0]/ori_image.shape[1] * new_w)
        #         y1 = int((pts[i - 1][1] - crop_y)/(ori_image.shape[0]-crop_y) * new_h)
        #         x2 = int(pts[i][0]/ori_image.shape[1] * new_w)
        #         y2 = int((pts[i][1] - crop_y)/(ori_image.shape[0]-crop_y) * new_h)

        #         cv2.line(draw_img, (x1,y1),
        #                  (x2,y2), (0,125,45), thickness=4)
        
        # plt.figure(dpi=720)
        # plt.imshow(draw_img, cmap='BrBG_r')
        # save files
        image_name = 'VIL100' + "_" + split_name + str(n_0)+".png"
        mask_name = 'VIL100' + "_" + split_name + str(n_0)+".png"
        txt_name = 'VIL100' + "_" + split_name + str(n_0)+".lines.txt"
        cv2.imwrite(os.path.join(img_path, image_name), img_new)
        cv2.imwrite(os.path.join(mask_path, mask_name), mas_new)
        save_txt_path = os.path.join(text_path, txt_name)
        with open(save_txt_path, 'w') as fp:
            lines = ''
            for pts in(points_list):
                line = ''
                for i in range(len(pts)):
                    x = pts[i][0]/ori_image.shape[1] * new_w
                    y = (pts[i][1] - crop_y)/(ori_image.shape[0]-crop_y) * new_h
                    line += f"{x:.2f} {y:.2f} "
                line+='\n'
            lines+=line
            #print(lines)
            fp.writelines(lines)
        n_0 = n_0 + 1