Created by: stuhailiu

从github上下载deeplabv3_xception65_bn的权重参数，直接用paddleseg加载进行预测，使用默认的参数配置，输出的miou只有0.59

[EVAL]#image=500 acc=0.9114 IoU=0.5917 [EVAL]Category IoU: [0.9464 0.6721 0.8490 0.4693 0.4214 0.2727 0.3392 0.4761 0.8506 0.5234 0.8905 0.5392 0.2792 0.8313 0.6819 0.7244 0.6624 0.2800 0.5329] [EVAL]Category Acc: [0.9841 0.8093 0.9030 0.7519 0.6044 0.5073 0.6370 0.7179 0.8937 0.7346 0.9478 0.7009 0.5528 0.8958 0.7783 0.8263 0.8320 0.6636 0.6784]

参数配置如下 TRAIN_CROP_SIZE: (512, 512) # (width, height), for unpadding rangescaling and stepscaling EVAL_CROP_SIZE: (512, 512) # (width, height), for unpadding rangescaling and stepscaling AUG: AUG_METHOD: "unpadding" # choice unpadding rangescaling and stepscaling FIX_RESIZE_SIZE: (512, 512) # (width, height), for unpadding

INF_RESIZE_VALUE: 500  # for rangescaling
MAX_RESIZE_VALUE: 600  # for rangescaling
MIN_RESIZE_VALUE: 400  # for rangescaling

MAX_SCALE_FACTOR: 1.25  # for stepscaling
MIN_SCALE_FACTOR: 0.75  # for stepscaling
SCALE_STEP_SIZE: 0.25  # for stepscaling
MIRROR: True

BATCH_SIZE: 4 DATASET: DATA_DIR: "E:/dakeai/cityscape" IMAGE_TYPE: "rgb" # choice rgb or rgba NUM_CLASSES: 19 TEST_FILE_LIST: "E:/dakeai/cityscape/list/cityscapes_val_list.txt" TRAIN_FILE_LIST: "E:/dakeai/cityscape/list/cityscapes_train_list.txt" VAL_FILE_LIST: "E:/dakeai/cityscape/list/cityscapes_val_list.txt" VIS_FILE_LIST: "E:/dakeai/cityscape/list/cityscapes_val_list.txt" IGNORE_INDEX: 255 SEPARATOR: " " FREEZE: MODEL_FILENAME: "model" PARAMS_FILENAME: "params" MODEL: MODEL_NAME: "deeplabv3p" DEFAULT_NORM_TYPE: "bn" DEEPLAB: BACKBONE: "xception_65" TRAIN: PRETRAINED_MODEL_DIR: "./pretrained_model/deeplabv3p_xception65_bn_coco/" MODEL_SAVE_DIR: "./saved_model/deeplabv3p_xception65_bn_pet/" SNAPSHOT_EPOCH: 10 TEST: TEST_MODEL: "./saved_model/deeplabv3p_xception65_bn_pet/final" SOLVER: NUM_EPOCHS: 100 LR: 0.005 LR_POLICY: "poly" OPTIMIZER: "sgd"