File size: 2,586 Bytes
d84c647 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 | from mmseg.apis import inference_segmentor, init_segmentor
import mmcv
import glob
import os
from tqdm import tqdm
import argparse
from PIL import Image
import numpy as np
from skimage.io import imsave
import pdb
def visualize_segmentation(img, seg_result, alpha = 0.4):
seg_color = np.zeros((img.shape))
seg_color[seg_result == 0] = (0, 0, 0) # background
seg_color[seg_result == 1] = (255, 0, 0) # left_hand
seg_color[seg_result == 2] = (0, 0, 255) # right_hand
seg_color[seg_result == 3] = (255, 0, 255) # left_object1
seg_color[seg_result == 4] = (0, 255, 255) # right_object1
seg_color[seg_result == 5] = (0, 255, 0) # two_object1
seg_color[seg_result == 6] = (255, 204, 255) # left_object2
seg_color[seg_result == 7] = (204, 255, 255) # right_object2
seg_color[seg_result == 8] = (204, 255, 204) # two_object2
vis = img * (1 - alpha) + seg_color * alpha
return vis
parser = argparse.ArgumentParser(description="")
parser.add_argument("--config_file", default='./work_dirs/upernet_swin_base_patch4_window12_512x512_160k_egohos_handobj2_pretrain_480x360_22K/upernet_swin_base_patch4_window12_512x512_160k_egohos_handobj2_pretrain_480x360_22K.py', type=str)
parser.add_argument("--checkpoint_file", default='./work_dirs/upernet_swin_base_patch4_window12_512x512_160k_egohos_handobj2_pretrain_480x360_22K/best_mIoU_iter_42000.pth', type=str)
parser.add_argument("--img_dir", default='../data/train/image', type=str)
parser.add_argument("--vis_dir", default='./work_dirs/upernet_swin_base_patch4_window12_512x512_160k_egohos_handobj2_pretrain_480x360_22K/outputs/train', type=str)
parser.add_argument("--seg_dir", default='./work_dirs/upernet_swin_base_patch4_window12_512x512_160k_egohos_handobj2_pretrain_480x360_22K/outputs/train_seg', type=str)
args = parser.parse_args()
# build the model from a config file and a checkpoint file
model = init_segmentor(args.config_file, args.checkpoint_file, device='cuda:0')
os.makedirs(args.vis_dir, exist_ok = True)
os.makedirs(args.seg_dir, exist_ok = True)
alpha = 0.3
for file in tqdm(glob.glob(args.img_dir + '/*')):
fname = os.path.basename(file).split('.')[0]
img = np.array(Image.open(os.path.join(args.img_dir, fname + '.jpg')))
seg_result = inference_segmentor(model, file)[0]
vis = visualize_segmentation(img, seg_result, alpha = alpha)
imsave(os.path.join(args.vis_dir, fname + '.jpg'), vis)
imsave(os.path.join(args.seg_dir, fname + '.png'), seg_result.astype(np.uint8))
# pdb.set_trace()
|