v0.5.0 (#124)

fixes https://github.com/zhanghang1989/PyTorch-Encoding/issues/118
fixes https://github.com/zhanghang1989/PyTorch-Encoding/issues/111
fixes https://github.com/zhanghang1989/PyTorch-Encoding/issues/84
fixes https://github.com/zhanghang1989/PyTorch-Encoding/issues/67
fixes https://github.com/zhanghang1989/PyTorch-Encoding/issues/88

encoding/lib/gpu/setup.py

@@ -7,8 +7,10 @@ setup(
         CUDAExtension('enclib_gpu', [
             'operator.cpp',
             'encoding_kernel.cu',
+            'encodingv2_kernel.cu',
             'syncbn_kernel.cu',
             'roi_align_kernel.cu',
+            'nms_kernel.cu',
             ]),
     ],
     cmdclass={

encoding/lib/gpu/syncbn_kernel.cu

-#include <ATen/ATen.h>
 #include <vector>
+#include <ATen/ATen.h>
+#include <ATen/cuda/CUDAContext.h>
 
 #include "common.h"
 #include "device_tensor.h"
@@ -180,7 +181,7 @@ at::Tensor BatchNorm_Forward_CUDA(
     const at::Tensor gamma_,
     const at::Tensor beta_) {
   auto output_ = at::zeros_like(input_);
-  cudaStream_t stream = at::globalContext().getCurrentCUDAStream();
+  cudaStream_t stream = at::cuda::getCurrentCUDAStream();
   dim3 blocks(input_.size(1));
   dim3 threads(getNumThreads(input_.size(2)));
   AT_DISPATCH_FLOATING_TYPES(input_.type(), "BatchNorm_Forward_CUDA", ([&] {
@@ -214,7 +215,7 @@ std::vector<at::Tensor> BatchNorm_Backward_CUDA(
   at::Tensor gradMean_ = at::zeros_like(mean_);
   at::Tensor gradStd_ = at::zeros_like(std_);
   /* cuda utils*/
-  cudaStream_t stream = at::globalContext().getCurrentCUDAStream();
+  cudaStream_t stream = at::cuda::getCurrentCUDAStream();
   dim3 blocks(input_.size(1));
   dim3 threads(getNumThreads(input_.size(2)));
   AT_DISPATCH_FLOATING_TYPES(input_.type(), "BatchNorm_Backward_CUDA", ([&] {
@@ -246,10 +247,10 @@ std::vector<at::Tensor> Sum_Square_Forward_CUDA(
   at::Tensor sum_ = input_.type().tensor({input_.size(1)}).zero_();
   at::Tensor square_ = input_.type().tensor({input_.size(1)}).zero_();
   /* cuda utils*/
-  cudaStream_t stream = at::globalContext().getCurrentCUDAStream();
+  cudaStream_t stream = at::cuda::getCurrentCUDAStream();
   dim3 blocks(input_.size(1));
   dim3 threads(getNumThreads(input_.size(2)));
-  AT_DISPATCH_FLOATING_TYPES(input_.type(), "BatchNorm_Backward_CUDA", ([&] {
+  AT_DISPATCH_FLOATING_TYPES(input_.type(), "SumSquare_forward_CUDA", ([&] {
     /* Device tensors */
     DeviceTensor<scalar_t, 3> input = devicetensor<scalar_t, 3>(input_);
     DeviceTensor<scalar_t, 1> sum = devicetensor<scalar_t, 1>(sum_);
@@ -269,10 +270,10 @@ at::Tensor Sum_Square_Backward_CUDA(
   /* outputs */
   at::Tensor gradInput_ = at::zeros_like(input_);
   /* cuda utils*/
-  cudaStream_t stream = at::globalContext().getCurrentCUDAStream();
+  cudaStream_t stream = at::cuda::getCurrentCUDAStream();
   dim3 blocks(input_.size(1));
   dim3 threads(getNumThreads(input_.size(2)));
-  AT_DISPATCH_FLOATING_TYPES(input_.type(), "BatchNorm_Backward_CUDA", ([&] {
+  AT_DISPATCH_FLOATING_TYPES(input_.type(), "SumSquare_Backward_CUDA", ([&] {
     /* Device tensors */
     DeviceTensor<scalar_t, 3> gradInput = devicetensor<scalar_t, 3>(gradInput_);
     DeviceTensor<scalar_t, 3> input = devicetensor<scalar_t, 3>(input_);

encoding/models/base.py

@@ -24,7 +24,7 @@ __all__ = ['BaseNet', 'MultiEvalModule']
 
 class BaseNet(nn.Module):
     def __init__(self, nclass, backbone, aux, se_loss, dilated=True, norm_layer=None,
-                 base_size=576, crop_size=608, mean=[.485, .456, .406],
+                 base_size=520, crop_size=480, mean=[.485, .456, .406],
                  std=[.229, .224, .225], root='~/.encoding/models'):
         super(BaseNet, self).__init__()
         self.nclass = nclass
@@ -99,6 +99,8 @@ class MultiEvalModule(DataParallel):
         elif len(kwargs) < len(inputs):
             kwargs.extend([{} for _ in range(len(inputs) - len(kwargs))])
         outputs = self.parallel_apply(replicas, inputs, kwargs)
+        #for out in outputs:
+        #    print('out.size()', out.size())
         return outputs
 
     def forward(self, image):

encoding/models/encnet.py

@@ -14,7 +14,8 @@ from .base import BaseNet
 from .fcn import FCNHead
 
 __all__ = ['EncNet', 'EncModule', 'get_encnet', 'get_encnet_resnet50_pcontext',
-           'get_encnet_resnet101_pcontext', 'get_encnet_resnet50_ade']
+           'get_encnet_resnet101_pcontext', 'get_encnet_resnet50_ade',
+           'get_encnet_resnet101_ade']
 
 class EncNet(BaseNet):
     def __init__(self, nclass, backbone, aux=True, se_loss=True, lateral=False,
@@ -43,8 +44,6 @@ class EncNet(BaseNet):
 class EncModule(nn.Module):
     def __init__(self, in_channels, nclass, ncodes=32, se_loss=True, norm_layer=None):
         super(EncModule, self).__init__()
-        #norm_layer = nn.BatchNorm1d if isinstance(norm_layer, nn.BatchNorm2d) else \
-        #    encoding.nn.BatchNorm1d
         self.se_loss = se_loss
         self.encoding = nn.Sequential(
             nn.Conv2d(in_channels, in_channels, 1, bias=False),
@@ -140,9 +139,9 @@ def get_encnet(dataset='pascal_voc', backbone='resnet50', pretrained=False,
         'ade20k': 'ade',
         'pcontext': 'pcontext',
     }
-    kwargs['lateral'] = True if dataset.lower() == 'pcontext' else False
+    kwargs['lateral'] = True if dataset.lower().startswith('p') else False
     # infer number of classes
-    from ..datasets import datasets, VOCSegmentation, VOCAugSegmentation, ADE20KSegmentation
+    from ..datasets import datasets
     model = EncNet(datasets[dataset.lower()].NUM_CLASS, backbone=backbone, root=root, **kwargs)
     if pretrained:
         from .model_store import get_model_file
@@ -167,7 +166,8 @@ def get_encnet_resnet50_pcontext(pretrained=False, root='~/.encoding/models', **
     >>> model = get_encnet_resnet50_pcontext(pretrained=True)
     >>> print(model)
     """
-    return get_encnet('pcontext', 'resnet50', pretrained, root=root, aux=True, **kwargs)
+    return get_encnet('pcontext', 'resnet50', pretrained, root=root, aux=True,
+                      base_size=520, crop_size=480, **kwargs)
 
 def get_encnet_resnet101_pcontext(pretrained=False, root='~/.encoding/models', **kwargs):
     r"""EncNet-PSP model from the paper `"Context Encoding for Semantic Segmentation"
@@ -186,7 +186,8 @@ def get_encnet_resnet101_pcontext(pretrained=False, root='~/.encoding/models', *
     >>> model = get_encnet_resnet101_pcontext(pretrained=True)
     >>> print(model)
     """
-    return get_encnet('pcontext', 'resnet101', pretrained, root=root, aux=True, **kwargs)
+    return get_encnet('pcontext', 'resnet101', pretrained, root=root, aux=True,
+                      base_size=520, crop_size=480, **kwargs)
 
 def get_encnet_resnet50_ade(pretrained=False, root='~/.encoding/models', **kwargs):
     r"""EncNet-PSP model from the paper `"Context Encoding for Semantic Segmentation"
@@ -205,4 +206,45 @@ def get_encnet_resnet50_ade(pretrained=False, root='~/.encoding/models', **kwarg
     >>> model = get_encnet_resnet50_ade(pretrained=True)
     >>> print(model)
     """
-    return get_encnet('ade20k', 'resnet50', pretrained, root=root, aux=True, **kwargs)
+    return get_encnet('ade20k', 'resnet50', pretrained, root=root, aux=True,
+                      base_size=520, crop_size=480, **kwargs)
+
+def get_encnet_resnet101_ade(pretrained=False, root='~/.encoding/models', **kwargs):
+    r"""EncNet-PSP model from the paper `"Context Encoding for Semantic Segmentation"
+    <https://arxiv.org/pdf/1803.08904.pdf>`_
+
+    Parameters
+    ----------
+    pretrained : bool, default False
+        Whether to load the pretrained weights for model.
+    root : str, default '~/.encoding/models'
+        Location for keeping the model parameters.
+
+
+    Examples
+    --------
+    >>> model = get_encnet_resnet50_ade(pretrained=True)
+    >>> print(model)
+    """
+    return get_encnet('ade20k', 'resnet101', pretrained, root=root, aux=True,
+                      base_size=640, crop_size=576, **kwargs)
+
+def get_encnet_resnet152_ade(pretrained=False, root='~/.encoding/models', **kwargs):
+    r"""EncNet-PSP model from the paper `"Context Encoding for Semantic Segmentation"
+    <https://arxiv.org/pdf/1803.08904.pdf>`_
+
+    Parameters
+    ----------
+    pretrained : bool, default False
+        Whether to load the pretrained weights for model.
+    root : str, default '~/.encoding/models'
+        Location for keeping the model parameters.
+
+
+    Examples
+    --------
+    >>> model = get_encnet_resnet50_ade(pretrained=True)
+    >>> print(model)
+    """
+    return get_encnet('ade20k', 'resnet152', pretrained, root=root, aux=True,
+                      base_size=520, crop_size=480, **kwargs)

encoding/models/fcn.py

@@ -101,8 +101,7 @@ def get_fcn(dataset='pascal_voc', backbone='resnet50', pretrained=False,
     if pretrained:
         from .model_store import get_model_file
         model.load_state_dict(torch.load(
-            get_model_file('fcn_%s_%s'%(backbone, acronyms[dataset]), root=root)),
-            strict= False)
+            get_model_file('fcn_%s_%s'%(backbone, acronyms[dataset]), root=root)))
     return model
 
 def get_fcn_resnet50_pcontext(pretrained=False, root='~/.encoding/models', **kwargs):

encoding/models/model_store.py

@@ -7,15 +7,19 @@ import zipfile
 from ..utils import download, check_sha1
 
 _model_sha1 = {name: checksum for checksum, name in [
-    ('853f2fb07aeb2927f7696e166b215609a987fd44', 'resnet50'),
-    ('5be5422ad7cb6a2e5f5a54070d0aa9affe69a9a4', 'resnet101'),
-    ('6cb047cda851de6aa31963e779fae5f4c299056a', 'deepten_minc'),
+    ('ebb6acbbd1d1c90b7f446ae59d30bf70c74febc1', 'resnet50'),
+    ('2a57e44de9c853fa015b172309a1ee7e2d0e4e2a', 'resnet101'),
+    ('0d43d698c66aceaa2bc0309f55efdd7ff4b143af', 'resnet152'),
+    ('2e22611a7f3992ebdee6726af169991bc26d7363', 'deepten_minc'),
     ('fc8c0b795abf0133700c2d4265d2f9edab7eb6cc', 'fcn_resnet50_ade'),
     ('eeed8e582f0fdccdba8579e7490570adc6d85c7c', 'fcn_resnet50_pcontext'),
     ('54f70c772505064e30efd1ddd3a14e1759faa363', 'psp_resnet50_ade'),
-    ('558e8904e123813f23dc0347acba85224650fe5f', 'encnet_resnet50_ade'),
-    ('7846a2f065e90ce70d268ba8ada1a92251587734', 'encnet_resnet50_pcontext'),
-    ('6f7c372259988bc2b6d7fc0007182e7835c31a11', 'encnet_resnet101_pcontext'),
+    ('075195c5237b778c718fd73ceddfa1376c18dfd0', 'deeplab_resnet50_ade'),
+    ('5ee47ee28b480cc781a195d13b5806d5bbc616bf', 'encnet_resnet101_coco'),
+    ('4de91d5922d4d3264f678b663f874da72e82db00', 'encnet_resnet50_pcontext'),
+    ('9f27ea13d514d7010e59988341bcbd4140fcc33d', 'encnet_resnet101_pcontext'),
+    ('07ac287cd77e53ea583f37454e17d30ce1509a4a', 'encnet_resnet50_ade'),
+    ('3f54fa3b67bac7619cd9b3673f5c8227cf8f4718', 'encnet_resnet101_ade'),
     ]}
 
 encoding_repo_url = 'https://hangzh.s3.amazonaws.com/'
@@ -52,9 +56,10 @@ def get_model_file(name, root=os.path.join('~', '.encoding', 'models')):
         if check_sha1(file_path, sha1_hash):
             return file_path
         else:
-            print('Mismatch in the content of model file detected. Downloading again.')
+            print('Mismatch in the content of model file {} detected.' +
+                  ' Downloading again.'.format(file_path))
     else:
-        print('Model file is not found. Downloading.')
+        print('Model file {} is not found. Downloading.'.format(file_path))
 
     if not os.path.exists(root):
         os.makedirs(root)

encoding/models/model_zoo.py

@@ -3,6 +3,7 @@
 from .fcn import *
 from .psp import *
 from .encnet import *
+from .deeplab import *
 
 __all__ = ['get_model']
 
@@ -29,8 +30,10 @@ def get_model(name, **kwargs):
         'encnet_resnet50_pcontext': get_encnet_resnet50_pcontext,
         'encnet_resnet101_pcontext': get_encnet_resnet101_pcontext,
         'encnet_resnet50_ade': get_encnet_resnet50_ade,
+        'encnet_resnet101_ade': get_encnet_resnet101_ade,
         'fcn_resnet50_ade': get_fcn_resnet50_ade,
         'psp_resnet50_ade': get_psp_resnet50_ade,
+        'deeplab_resnet50_ade': get_deeplab_resnet50_ade,
         }
     name = name.lower()
     if name not in models:

encoding/models/psp.py

@@ -58,7 +58,7 @@ def get_psp(dataset='pascal_voc', backbone='resnet50', pretrained=False,
         'ade20k': 'ade',
     }
     # infer number of classes
-    from ..datasets import datasets, VOCSegmentation, VOCAugSegmentation, ADE20KSegmentation
+    from ..datasets import datasets
     model = PSP(datasets[dataset.lower()].NUM_CLASS, backbone=backbone, root=root, **kwargs)
     if pretrained:
         from .model_store import get_model_file

encoding/nn/customize.py

@@ -40,7 +40,7 @@ def softmax_crossentropy(input, target, weight, size_average, ignore_index, redu
 class SegmentationLosses(CrossEntropyLoss):
     """2D Cross Entropy Loss with Auxilary Loss"""
     def __init__(self, se_loss=False, se_weight=0.2, nclass=-1,
-                 aux=False, aux_weight=0.2, weight=None,
+                 aux=False, aux_weight=0.4, weight=None,
                  size_average=True, ignore_index=-1):
         super(SegmentationLosses, self).__init__(weight, size_average, ignore_index)
         self.se_loss = se_loss
@@ -62,14 +62,14 @@ class SegmentationLosses(CrossEntropyLoss):
             pred, se_pred, target = tuple(inputs)
             se_target = self._get_batch_label_vector(target, nclass=self.nclass).type_as(pred)
             loss1 = super(SegmentationLosses, self).forward(pred, target)
-            loss2 = self.bceloss(F.sigmoid(se_pred), se_target)
+            loss2 = self.bceloss(torch.sigmoid(se_pred), se_target)
             return loss1 + self.se_weight * loss2
         else:
             pred1, se_pred, pred2, target = tuple(inputs)
             se_target = self._get_batch_label_vector(target, nclass=self.nclass).type_as(pred1)
             loss1 = super(SegmentationLosses, self).forward(pred1, target)
             loss2 = super(SegmentationLosses, self).forward(pred2, target)
-            loss3 = self.bceloss(F.sigmoid(se_pred), se_target)
+            loss3 = self.bceloss(torch.sigmoid(se_pred), se_target)
             return loss1 + self.aux_weight * loss2 + self.se_weight * loss3
 
     @staticmethod

encoding/nn/encoding.py

@@ -15,7 +15,7 @@ import torch.nn.functional as F
 from torch.autograd import Variable
 from torch.nn.modules.utils import _pair
 
-from ..functions import scaledL2, aggregate, pairwise_cosine
+from ..functions import scaled_l2, aggregate, pairwise_cosine
 
 __all__ = ['Encoding', 'EncodingDrop', 'Inspiration', 'UpsampleConv2d']
 
@@ -90,18 +90,17 @@ class Encoding(Module):
     def forward(self, X):
         # input X is a 4D tensor
         assert(X.size(1) == self.D)
+        B, D = X.size(0), self.D
         if X.dim() == 3:
-            # BxDxN
-            B, D = X.size(0), self.D
+            # BxDxN => BxNxD
             X = X.transpose(1, 2).contiguous()
         elif X.dim() == 4:
-            # BxDxHxW
-            B, D = X.size(0), self.D
+            # BxDxHxW => Bx(HW)xD
             X = X.view(B, D, -1).transpose(1, 2).contiguous()
         else:
             raise RuntimeError('Encoding Layer unknown input dims!')
-        # assignment weights NxKxD
-        A = F.softmax(scaledL2(X, self.codewords, self.scale), dim=1)
+        # assignment weights BxNxK
+        A = F.softmax(scaled_l2(X, self.codewords, self.scale), dim=2)
         # aggregate
         E = aggregate(A, X, self.codewords)
         return E
@@ -149,7 +148,7 @@ class EncodingDrop(Module):
             raise RuntimeError('Encoding Layer unknown input dims!')
         self._drop()
         # assignment weights
-        A = F.softmax(scaledL2(X, self.codewords, self.scale), dim=1)
+        A = F.softmax(scaled_l2(X, self.codewords, self.scale), dim=2)
         # aggregate
         E = aggregate(A, X, self.codewords)
         self._drop()

encoding/utils/__init__.py

@@ -10,7 +10,7 @@
 
 """Encoding Util Tools"""
 from .lr_scheduler import LR_Scheduler
-from .metrics import batch_intersection_union, batch_pix_accuracy
+from .metrics import SegmentationMetric, batch_intersection_union, batch_pix_accuracy
 from .pallete import get_mask_pallete
 from .train_helper import get_selabel_vector, EMA
 from .presets import load_image

encoding/utils/metrics.py

@@ -8,18 +8,70 @@
 ## LICENSE file in the root directory of this source tree
 ##+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 
+import threading
 import numpy as np
 import torch
 
-def batch_pix_accuracy(predict, target):
+class SegmentationMetric(object):
+    """Computes pixAcc and mIoU metric scroes
+    """
+    def __init__(self, nclass):
+        self.nclass = nclass
+        self.lock = threading.Lock()
+        self.reset()
+
+    def update(self, labels, preds):
+        def evaluate_worker(self, label, pred):
+            correct, labeled = batch_pix_accuracy(
+                pred, label)
+            inter, union = batch_intersection_union(
+                pred, label, self.nclass)
+            with self.lock:
+                self.total_correct += correct
+                self.total_label += labeled
+                self.total_inter += inter
+                self.total_union += union
+            return
+
+        if isinstance(preds, torch.Tensor):
+            evaluate_worker(self, labels, preds)
+        elif isinstance(preds, (list, tuple)):
+            threads = [threading.Thread(target=evaluate_worker,
+                                        args=(self, label, pred),
+                                       )
+                       for (label, pred) in zip(labels, preds)]
+            for thread in threads:
+                thread.start()
+            for thread in threads:
+                thread.join()
+        else:
+            raise NotImplemented
+
+    def get(self):
+        pixAcc = 1.0 * self.total_correct / (np.spacing(1) + self.total_label)
+        IoU = 1.0 * self.total_inter / (np.spacing(1) + self.total_union)
+        mIoU = IoU.mean()
+        return pixAcc, mIoU
+ 
+    def reset(self):
+        self.total_inter = 0
+        self.total_union = 0
+        self.total_correct = 0
+        self.total_label = 0
+        return
+
+
+def batch_pix_accuracy(output, target):
     """Batch Pixel Accuracy
     Args:
         predict: input 4D tensor
         target: label 3D tensor
     """
-    _, predict = torch.max(predict, 1)
-    predict = predict.cpu().numpy() + 1
-    target = target.cpu().numpy() + 1
+    _, predict = torch.max(output, 1)
+
+    predict = predict.cpu().numpy().astype('int64') + 1
+    target = target.cpu().numpy().astype('int64') + 1
+
     pixel_labeled = np.sum(target > 0)
     pixel_correct = np.sum((predict == target)*(target > 0))
     assert pixel_correct <= pixel_labeled, \
@@ -27,19 +79,19 @@ def batch_pix_accuracy(predict, target):
     return pixel_correct, pixel_labeled
 
 
-def batch_intersection_union(predict, target, nclass):
+def batch_intersection_union(output, target, nclass):
     """Batch Intersection of Union
     Args:
         predict: input 4D tensor
         target: label 3D tensor
         nclass: number of categories (int)
     """
-    _, predict = torch.max(predict, 1)
+    _, predict = torch.max(output, 1)
     mini = 1
     maxi = nclass
     nbins = nclass
-    predict = predict.cpu().numpy() + 1
-    target = target.cpu().numpy() + 1
+    predict = predict.cpu().numpy().astype('int64') + 1
+    target = target.cpu().numpy().astype('int64') + 1
 
     predict = predict * (target > 0).astype(predict.dtype)
     intersection = predict * (predict == target)

encoding/utils/pallete.py

@@ -21,11 +21,10 @@ def get_mask_pallete(npimg, dataset='detail'):
         out_img.putpalette(adepallete)
     elif dataset == 'cityscapes':
         out_img.putpalette(citypallete)
-    else:
+    elif dataset in ('detail', 'pascal_voc', 'pascal_aug'):
         out_img.putpalette(vocpallete)
     return out_img
 
-
 def _get_voc_pallete(num_cls):
     n = num_cls
     pallete = [0]*(n*3)

experiments/recognition/dataset/minc.py

@@ -94,9 +94,9 @@ class Dataloader():
             normalize,
         ])
 
-        trainset = MINCDataloder(root=os.path.expanduser('~/data/minc-2500/'), 
+        trainset = MINCDataloder(root=os.path.expanduser('~/.encoding/data/minc-2500/'), 
             train=True, transform=transform_train)
-        testset = MINCDataloder(root=os.path.expanduser('~/data/minc-2500/'), 
+        testset = MINCDataloder(root=os.path.expanduser('~/.encoding/data/minc-2500/'), 
             train=False, transform=transform_test)
     
         kwargs = {'num_workers': 8, 'pin_memory': True} if args.cuda else {}

experiments/recognition/main.py

@@ -96,7 +96,7 @@ def main():
 
             train_loss += loss.data.item()
             pred = output.data.max(1)[1] 
-            correct += pred.eq(target.data).cpu().sum()
+            correct += pred.eq(target.data).cpu().sum().item()
             total += target.size(0)
             err = 100.0 - 100.0 * correct / total
             tbar.set_description('\rLoss: %.3f | Err: %.3f%% (%d/%d)' % \

experiments/recognition/model/download_models.py

@@ -2,4 +2,4 @@ import encoding
 import shutil
 
 encoding.models.get_model_file('deepten_minc', root='./')
-shutil.move('deepten_minc-6cb047cd.pth', 'deepten_minc.pth')
+shutil.move('deepten_minc-2e22611a.pth', 'deepten_minc.pth')

experiments/recognition/model/mynn.py

@@ -86,6 +86,46 @@ class Bottleneck(nn.Module):
         return residual + self.conv_block(x)
         
 ##+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+class EncLayerV2(nn.Module):
+    def __init__(self, channel, K=16, reduction=4):
+        super(EncLayerV2, self).__init__()
+        out_channel = int(channel / reduction)
+        self.fc = nn.Sequential(
+            nn.Conv2d(channel, out_channel, 1),
+            nn.BatchNorm2d(out_channel),
+            nn.ReLU(inplace=True),
+            encoding.nn.EncodingV2(D=out_channel,K=K),
+            encoding.nn.View(-1, out_channel*K),
+            encoding.nn.Normalize(),
+            nn.Linear(out_channel*K, channel),
+            nn.Sigmoid()
+        )
+
+    def forward(self, x):
+        b, c, _, _ = x.size()
+        y = self.fc(x).view(b, c, 1, 1)
+        return x * y
+
+class EncLayerV3(nn.Module):
+    def __init__(self, channel, K=16, reduction=4):
+        super(EncLayerV3, self).__init__()
+        out_channel = int(channel / reduction)
+        self.fc = nn.Sequential(
+            nn.Conv2d(channel, out_channel, 1),
+            nn.BatchNorm2d(out_channel),
+            nn.ReLU(inplace=True),
+            encoding.nn.EncodingV3(D=out_channel,K=K),
+            encoding.nn.View(-1, out_channel*K),
+            encoding.nn.Normalize(),
+            nn.Linear(out_channel*K, channel),
+            nn.Sigmoid()
+        )
+
+    def forward(self, x):
+        b, c, _, _ = x.size()
+        y = self.fc(x).view(b, c, 1, 1)
+        return x * y
+
 class EncLayer(nn.Module):
     def __init__(self, channel, K=16, reduction=4):
         super(EncLayer, self).__init__()

experiments/recognition/option.py

@@ -40,8 +40,8 @@ class Options():
         # lr setting
         parser.add_argument('--lr', type=float, default=0.1, metavar='LR',
             help='learning rate (default: 0.1)')
-        parser.add_argument('--lr-scheduler', type=str, default='step', 
-            help='learning rate scheduler (default: step)')
+        parser.add_argument('--lr-scheduler', type=str, default='cos', 
+            help='learning rate scheduler (default: cos)')
         parser.add_argument('--lr-step', type=int, default=40, metavar='LR',
             help='learning rate step (default: 40)')
         # optimizer

experiments/segmentation/option.py

@@ -25,15 +25,21 @@ class Options():
                             $(HOME)/data)')
         parser.add_argument('--workers', type=int, default=16,
                             metavar='N', help='dataloader threads')
-        parser.add_argument('--base-size', type=int, default=608,
+        parser.add_argument('--base-size', type=int, default=520,
                             help='base image size')
-        parser.add_argument('--crop-size', type=int, default=576,
+        parser.add_argument('--crop-size', type=int, default=480,
                             help='crop image size')
+        parser.add_argument('--train-split', type=str, default='train',
+                        help='dataset train split (default: train)')
         # training hyper params
         parser.add_argument('--aux', action='store_true', default= False,
                             help='Auxilary Loss')
+        parser.add_argument('--aux-weight', type=float, default=0.2,
+                            help='Auxilary loss weight (default: 0.2)')
         parser.add_argument('--se-loss', action='store_true', default= False,
                             help='Semantic Encoding Loss SE-loss')
+        parser.add_argument('--se-weight', type=float, default=0.2,
+                            help='SE-loss weight (default: 0.2)')
         parser.add_argument('--epochs', type=int, default=None, metavar='N',
                             help='number of epochs to train (default: auto)')
         parser.add_argument('--start_epoch', type=int, default=0,
@@ -68,12 +74,7 @@ class Options():
         # finetuning pre-trained models
         parser.add_argument('--ft', action='store_true', default= False,
                             help='finetuning on a different dataset')
-        parser.add_argument('--pre-class', type=int, default=None,
-                            help='num of pre-trained classes \
-                            (default: None)')
         # evaluation option
-        parser.add_argument('--ema', action='store_true', default= False,
-                            help='using EMA evaluation')
         parser.add_argument('--eval', action='store_true', default= False,
                             help='evaluating mIoU')
         parser.add_argument('--no-val', action='store_true', default= False,
@@ -90,10 +91,12 @@ class Options():
         # default settings for epochs, batch_size and lr
         if args.epochs is None:
             epoches = {
+                'coco': 30,
+                'citys': 180,
                 'pascal_voc': 50,
                 'pascal_aug': 50,
                 'pcontext': 80,
-                'ade20k': 160,
+                'ade20k': 120,
             }
             args.epochs = epoches[args.dataset.lower()]
         if args.batch_size is None:
@@ -102,10 +105,13 @@ class Options():
             args.test_batch_size = args.batch_size
         if args.lr is None:
             lrs = {
+                'coco': 0.01,
+                'citys': 0.01,
                 'pascal_voc': 0.0001,
                 'pascal_aug': 0.001,
                 'pcontext': 0.001,
                 'ade20k': 0.01,
             }
             args.lr = lrs[args.dataset.lower()] / 16 * args.batch_size
+        print(args)
         return args

experiments/segmentation/test.py

@@ -21,10 +21,6 @@ from encoding.models import get_model, get_segmentation_model, MultiEvalModule
 
 from option import Options
 
-torch_ver = torch.__version__[:3]
-if torch_ver == '0.3':
-    from torch.autograd import Variable
-
 def test(args):
     # output folder
     outdir = 'outdir'
@@ -64,58 +60,29 @@ def test(args):
         print("=> loaded checkpoint '{}' (epoch {})".format(args.resume, checkpoint['epoch']))
 
     print(model)
-    evaluator = MultiEvalModule(model, testset.num_class).cuda()
+    scales = [0.5, 0.75, 1.0, 1.25, 1.5, 1.75, 2.0, 2.25] if args.dataset == 'citys' else \
+        [0.5, 0.75, 1.0, 1.25, 1.5, 1.75]
+    evaluator = MultiEvalModule(model, testset.num_class, scales=scales).cuda()
     evaluator.eval()
+    metric = utils.SegmentationMetric(testset.num_class)
 
     tbar = tqdm(test_data)
-    def eval_batch(image, dst, evaluator, eval_mode):
-        if eval_mode:
-            # evaluation mode on validation set
-            targets = dst
-            outputs = evaluator.parallel_forward(image)
-            batch_inter, batch_union, batch_correct, batch_label = 0, 0, 0, 0
-            for output, target in zip(outputs, targets):
-                correct, labeled = utils.batch_pix_accuracy(output.data.cpu(), target)
-                inter, union = utils.batch_intersection_union(
-                    output.data.cpu(), target, testset.num_class)
-                batch_correct += correct
-                batch_label += labeled
-                batch_inter += inter
-                batch_union += union
-            return batch_correct, batch_label, batch_inter, batch_union
+    for i, (image, dst) in enumerate(tbar):
+        if args.eval:
+            with torch.no_grad():
+                predicts = evaluator.parallel_forward(image)
+                metric.update(dst, predicts)
+                pixAcc, mIoU = metric.get()
+                tbar.set_description( 'pixAcc: %.4f, mIoU: %.4f' % (pixAcc, mIoU))
         else:
-            # test mode, dump the results
-            im_paths = dst
-            outputs = evaluator.parallel_forward(image)
-            predicts = [torch.max(output, 1)[1].cpu().numpy() + testset.pred_offset
-                        for output in outputs]
-            for predict, impath in zip(predicts, im_paths):
+            with torch.no_grad():
+                outputs = evaluator.parallel_forward(image)
+                predicts = [testset.make_pred(torch.max(output, 1)[1].cpu().numpy())
+                            for output in outputs]
+            for predict, impath in zip(predicts, dst):
                 mask = utils.get_mask_pallete(predict, args.dataset)
                 outname = os.path.splitext(impath)[0] + '.png'
                 mask.save(os.path.join(outdir, outname))
-            # dummy outputs for compatible with eval mode
-            return 0, 0, 0, 0
-
-    total_inter, total_union, total_correct, total_label = \
-        np.int64(0), np.int64(0), np.int64(0), np.int64(0)
-    for i, (image, dst) in enumerate(tbar):
-        if torch_ver == "0.3":
-            image = Variable(image, volatile=True)
-            correct, labeled, inter, union = eval_batch(image, dst, evaluator, args.eval)
-        else:
-            with torch.no_grad():
-                correct, labeled, inter, union = eval_batch(image, dst, evaluator, args.eval)
-        if args.eval:
-            total_correct += correct.astype('int64')
-            total_label += labeled.astype('int64')
-            total_inter += inter.astype('int64')
-            total_union += union.astype('int64')
-            pixAcc = np.float64(1.0) * total_correct / (np.spacing(1, dtype=np.float64) + total_label)
-            IoU = np.float64(1.0) * total_inter / (np.spacing(1, dtype=np.float64) + total_union)
-            mIoU = IoU.mean()
-            tbar.set_description(
-                'pixAcc: %.4f, mIoU: %.4f' % (pixAcc, mIoU))
-
 
 if __name__ == "__main__":
     args = Options().parse()