Giters

VainF / DeepLabV3Plus-Pytorch

Pretrained DeepLabv3 and DeepLabv3+ for Pascal VOC & Cityscapes

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

extremely slow

TornadoooWang opened this issue · comments

commented

It's not any code error or configuration error, I just use cityscrapes dataset to predict pretrained model with only 1 img,it works extremely slow in those line below on my camputer,and the output is a grey img,which is obviously wrong.

                   pred = model(img).max(1)[1].gpu().numpy()[0] # HW
        colorized_preds = decode_fn(pred).astype('uint8')
        colorized_preds = Image.fromarray(colorized_preds)
        if opts.save_val_results_to:
            colorized_preds.save(os.path.join(opts.save_val_results_to, img_name+'.png'))
commented

use this command in readme

python predict.py --input datasets/data/cityscapes/leftImg8bit/train/bremen/bremen_000000_000019_leftImg8bit.png --dataset cityscapes --model deeplabv3plus_mobilenet --ckpt checkpoints/best_deeplabv3plus_mobilenet_cityscapes_os16.pth --save_val_results_to test_results

commented

I find the error ,because of the old version of torch etc.3060 GPU doesn't fit in.