Add files via upload

lib/config/__init__.py

@@ -0,0 +1,8 @@
+
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+from .default import _C as config
+from .default import update_config
+from .models import MODEL_EXTRAS

lib/config/default.py

@@ -0,0 +1,154 @@
+
+
+
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+import os
+
+from yacs.config import CfgNode as CN
+
+
+_C = CN()
+
+_C.OUTPUT_DIR = ''
+_C.LOG_DIR = ''
+_C.GPUS = (0,) #修改
+_C.WORKERS = 1
+_C.PRINT_FREQ = 10
+_C.AUTO_RESUME = False
+_C.PIN_MEMORY = True
+_C.RANK = 0
+_C.sub_model_index = 0
+
+# Cudnn related params
+_C.CUDNN = CN()
+_C.CUDNN.BENCHMARK = True
+_C.CUDNN.DETERMINISTIC = False
+_C.CUDNN.ENABLED = True
+
+# common params for NETWORK
+_C.MODEL = CN()
+_C.MODEL.NAME = 'seg_hrnet_without_interpolate '
+_C.MODEL.PRETRAINED = ''
+_C.MODEL.ALIGN_CORNERS = True
+_C.MODEL.NUM_OUTPUTS = 1
+_C.MODEL.EXTRA = CN(new_allowed=True)
+
+
+_C.MODEL.OCR = CN()
+_C.MODEL.OCR.MID_CHANNELS = 512
+_C.MODEL.OCR.KEY_CHANNELS = 256
+_C.MODEL.OCR.DROPOUT = 0.05
+_C.MODEL.OCR.SCALE = 1
+
+_C.LOSS = CN()
+_C.LOSS.USE_OHEM = False
+_C.LOSS.OHEMTHRES = 0.9
+_C.LOSS.OHEMKEEP = 100000
+_C.LOSS.CLASS_BALANCE = False
+_C.LOSS.BALANCE_WEIGHTS = [1]
+######## ACW_loss ########
+_C.LOSS.USE_ACW = False
+######## ACW_loss ########
+
+######## Attention ########
+_C.ATTENTION = CN()
+_C.ATTENTION.HSN_POSITION = '0'
+# 1, 2, 3, 1+2+3, 2+3
+_C.ATTENTION.PSNL_POSITION = '0'
+# 1, 2, 3, 1+2+3, 2+3
+_C.ATTENTION.ORDER = '0'
+# H, P, HP, PH, H/P
+######## Attention ########
+
+# DATASET related params
+_C.DATASET = CN()
+_C.DATASET.ROOT = ''
+_C.DATASET.DATASET = 'cityscapes'
+_C.DATASET.NUM_CLASSES = 11 #修改
+_C.DATASET.TRAIN_SET = 'list/cityscapes/train.lst'
+_C.DATASET.EXTRA_TRAIN_SET = ''
+_C.DATASET.TEST_SET = 'list/cityscapes/val.lst'
+
+# training
+_C.TRAIN = CN()
+
+_C.TRAIN.FREEZE_LAYERS = ''
+_C.TRAIN.FREEZE_EPOCHS = -1
+_C.TRAIN.NONBACKBONE_KEYWORDS = []
+_C.TRAIN.NONBACKBONE_MULT = 10
+
+_C.TRAIN.IMAGE_SIZE = [64, 64]  # width * height #修改
+_C.TRAIN.BASE_SIZE = 2048
+_C.TRAIN.DOWNSAMPLERATE = 1
+_C.TRAIN.FLIP = True
+_C.TRAIN.MULTI_SCALE = False
+_C.TRAIN.SCALE_FACTOR = 16
+
+_C.TRAIN.RANDOM_BRIGHTNESS = False
+_C.TRAIN.RANDOM_BRIGHTNESS_SHIFT_VALUE = 10
+
+_C.TRAIN.LR_FACTOR = 0.1
+_C.TRAIN.LR_STEP = [90, 110]
+_C.TRAIN.LR = 0.01
+_C.TRAIN.EXTRA_LR = 0.001
+
+_C.TRAIN.OPTIMIZER = 'sgd'
+_C.TRAIN.MOMENTUM = 0.9
+_C.TRAIN.WD = 0.0001
+_C.TRAIN.NESTEROV = False
+_C.TRAIN.IGNORE_LABEL = -1
+
+_C.TRAIN.BEGIN_EPOCH = 0
+_C.TRAIN.END_EPOCH = 1000
+_C.TRAIN.EXTRA_EPOCH = 0
+
+_C.TRAIN.RESUME = False
+
+_C.TRAIN.BATCH_SIZE_PER_GPU = 2
+_C.TRAIN.SHUFFLE = True
+# only using some training samples
+_C.TRAIN.NUM_SAMPLES = 0
+
+# testing
+_C.TEST = CN()
+
+_C.TEST.IMAGE_SIZE = [64, 64]  # width * height #修改
+_C.TEST.BASE_SIZE = 2048
+
+_C.TEST.BATCH_SIZE_PER_GPU = 1024
+# only testing some samples
+_C.TEST.NUM_SAMPLES = 0
+
+_C.TEST.MODEL_FILE = '/home/xieweiying/HRNet-attention_v2_64/tools/output/cityscapes/seg_hrnet_w48_train_512x1024_sgd_lr1e-2_wd5e-4_bs_12_epoch484/最原始模型90epochv264x64/best.pth' #修改 
+_C.TEST.FLIP_TEST = True
+_C.TEST.MULTI_SCALE = False
+_C.TEST.SCALE_LIST = [1]
+
+_C.TEST.OUTPUT_INDEX = -1
+
+# debug
+_C.DEBUG = CN()
+_C.DEBUG.DEBUG = False
+_C.DEBUG.SAVE_BATCH_IMAGES_GT = False
+_C.DEBUG.SAVE_BATCH_IMAGES_PRED = False
+_C.DEBUG.SAVE_HEATMAPS_GT = False
+_C.DEBUG.SAVE_HEATMAPS_PRED = False
+
+
+def update_config(cfg, args):
+    cfg.defrost()
+
+    cfg.merge_from_file(args.cfg)
+    cfg.merge_from_list(args.opts)
+
+    cfg.freeze()
+
+
+if __name__ == '__main__':
+    import sys
+    with open(sys.argv[1], 'w') as f:
+        print(_C, file=f)
+

lib/config/models.py

@@ -0,0 +1,42 @@
+
+
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+from yacs.config import CfgNode as CN
+
+# high_resoluton_net related params for segmentation
+HIGH_RESOLUTION_NET = CN()
+HIGH_RESOLUTION_NET.PRETRAINED_LAYERS = ['*']
+HIGH_RESOLUTION_NET.STEM_INPLANES = 64
+HIGH_RESOLUTION_NET.FINAL_CONV_KERNEL = 1
+HIGH_RESOLUTION_NET.WITH_HEAD = True
+
+HIGH_RESOLUTION_NET.STAGE2 = CN()
+HIGH_RESOLUTION_NET.STAGE2.NUM_MODULES = 1
+HIGH_RESOLUTION_NET.STAGE2.NUM_BRANCHES = 2
+HIGH_RESOLUTION_NET.STAGE2.NUM_BLOCKS = [4, 4]
+HIGH_RESOLUTION_NET.STAGE2.NUM_CHANNELS = [32, 64]
+HIGH_RESOLUTION_NET.STAGE2.BLOCK = 'BASIC'
+HIGH_RESOLUTION_NET.STAGE2.FUSE_METHOD = 'SUM'
+
+HIGH_RESOLUTION_NET.STAGE3 = CN()
+HIGH_RESOLUTION_NET.STAGE3.NUM_MODULES = 1
+HIGH_RESOLUTION_NET.STAGE3.NUM_BRANCHES = 3
+HIGH_RESOLUTION_NET.STAGE3.NUM_BLOCKS = [4, 4, 4]
+HIGH_RESOLUTION_NET.STAGE3.NUM_CHANNELS = [32, 64, 128]
+HIGH_RESOLUTION_NET.STAGE3.BLOCK = 'BASIC'
+HIGH_RESOLUTION_NET.STAGE3.FUSE_METHOD = 'SUM'
+
+HIGH_RESOLUTION_NET.STAGE4 = CN()
+HIGH_RESOLUTION_NET.STAGE4.NUM_MODULES = 1
+HIGH_RESOLUTION_NET.STAGE4.NUM_BRANCHES = 4
+HIGH_RESOLUTION_NET.STAGE4.NUM_BLOCKS = [4, 4, 4, 4]
+HIGH_RESOLUTION_NET.STAGE4.NUM_CHANNELS = [32, 64, 128, 256]
+HIGH_RESOLUTION_NET.STAGE4.BLOCK = 'BASIC'
+HIGH_RESOLUTION_NET.STAGE4.FUSE_METHOD = 'SUM'
+
+MODEL_EXTRAS = {
+    'seg_hrnet': HIGH_RESOLUTION_NET,
+}

lib/core/Fwiou.py

@@ -0,0 +1,126 @@
+import numpy as np
+from sklearn.metrics import confusion_matrix
+
+class _StreamMetrics(object):
+    def __init__(self):
+        """ Overridden by subclasses """
+        raise NotImplementedError()
+
+    def update(self, gt, pred):
+        """ Overridden by subclasses """
+        raise NotImplementedError()
+
+    def get_results(self):
+        """ Overridden by subclasses """
+        raise NotImplementedError()
+
+    def to_str(self, metrics):
+        """ Overridden by subclasses """
+        raise NotImplementedError()
+
+    def reset(self):
+        """ Overridden by subclasses """
+        raise NotImplementedError()      
+
+class StreamSegMetrics(_StreamMetrics):
+    """
+    Stream Metrics for Semantic Segmentation Task
+    """
+    def __init__(self, n_classes):
+        self.n_classes = n_classes
+        self.confusion_matrix = np.zeros((n_classes, n_classes))
+
+    def update(self, label_trues, label_preds):
+        for lt, lp in zip(label_trues, label_preds):
+            self.confusion_matrix += self._fast_hist( lt.flatten(), lp.flatten() )
+
+    @staticmethod
+    def to_str(results):
+        string = "\n"
+        for k, v in results.items():
+            if k!="Class IoU":
+                string += "%s: %f\n"%(k, v)
+            else:
+                for i in range(len(v)):
+                    string += "%s_%d: %f\n"%(k,i,v[i])
+                    #string += "%s_%d\n"%(k,i)
+        #string+='Class IoU:\n'
+        #for k, v in results['Class IoU'].items():
+        #    string += "\tclass %d: %f\n"%(k, v)
+        return string
+
+    def _fast_hist(self, label_true, label_pred):
+        mask = (label_true >= 0) & (label_true < self.n_classes)
+        hist = np.bincount(
+            self.n_classes * label_true[mask].astype(int) + label_pred[mask],
+            minlength=self.n_classes ** 2,
+        ).reshape(self.n_classes, self.n_classes)
+        return hist
+
+    def get_results(self):
+        """Returns accuracy score evaluation result.
+            - overall accuracy
+            - mean accuracy
+            - mean IU
+            - fwavacc
+
+        """
+        hist = self.confusion_matrix
+        acc = np.diag(hist).sum() / hist.sum()
+        acc_cls = np.diag(hist) / hist.sum(axis=1)
+        acc_cls = np.nanmean(acc_cls)
+        iu = np.diag(hist) / (hist.sum(axis=1) + hist.sum(axis=0) - np.diag(hist))
+        mean_iu = np.nanmean(iu)
+        freq = hist.sum(axis=1) / hist.sum()
+        freq[0]=0
+        freq[2]=0
+        freq[4]=0
+        freq[5]=0
+        freq[8]=0
+        freq[10]=0
+        freq = freq/freq.sum()
+        fwavacc = (freq[freq > 0] * iu[freq > 0]).sum()
+        #cls_iu = dict(zip(range(self.n_classes), iu))
+        cls_iu = []
+        for i in range(len(iu)):
+            iu_i = iu[i]
+            cls_iu.append(iu_i)
+        cls_iu = np.array(cls_iu)
+        return {
+                "Overall Acc": acc,
+                "Mean Acc": acc_cls,
+                "FreqW Acc": fwavacc,
+                "Mean IoU": mean_iu,
+                "Class IoU": cls_iu,
+            }
+
+    def reset(self):
+        self.confusion_matrix = np.zeros((self.n_classes, self.n_classes))
+
+class AverageMeter(object):
+    """Computes average values"""
+    def __init__(self):
+        self.book = dict()
+
+    def reset_all(self):
+        self.book.clear()
+
+    def reset(self, id):
+        item = self.book.get(id, None)
+        if item is not None:
+            item[0] = 0
+            item[1] = 0
+
+    def update(self, id, val):
+        record = self.book.get(id, None)
+        if record is None:
+            self.book[id] = [val, 1]
+        else:
+            record[0]+=val
+            record[1]+=1
+
+    def get_results(self, id):
+        record = self.book.get(id, None)
+        assert record is not None
+        return record[0] / record[1]
+