PSPNet_PyTorch
An implementation of PSPNet: Pyramid Scene Parsing Network, CVPR2017
Training
You can download pretrained model on COCO here for training on VOC2012 and pretrained model here for training on Cityscapes.
usage: train.py [-h] [--is-training] [--learning-rate LEARNING_RATE]
[--weight-decay WEIGHT_DECAY] [--momentum MOMENTUM]
[--power POWER] [--dataset DATASET] [--random-mirror]
[--random-scale] [--not-restore-last]
[--random-seed RANDOM_SEED] [--logFile LOGFILE] [--cuda CUDA]
[--gpus GPUS]
exmaple: train.py --dataset voc12
Evaluation (on val set)
usage: evaluate.py [-h] [--dataset DATASET] [--cuda CUDA] [--gpus GPUS]
Testing (on test set)
usage: test.py [-h] [--dataset DATASET] [--cuda CUDA] [--gpus GPUS]