merge eval_camvid and eval_cityscapes

chainercv/links/model/segnet/segnet_basic.py

@@ -52,7 +52,7 @@ class SegNetBasic(chainer.Chain):
         'camvid': {
             'param': {'n_class': 11},
             'url': 'https://chainercv-models.preferred.jp/'
-            'segnet_camvid_trained_2017_05_28.npz'
+            'segnet_camvid_trained_2018_12_05.npz'
         }
     }
 

examples/segnet/README.md

@@ -53,7 +53,7 @@ $ python demo.py [--gpu <gpu>] [--pretrained-model <model_path>] 0001TP_008550.p
 | ChainerCV      | 82.7 %          | **67.1 %**     | **49.4 %** |
 | Official       | **82.8 %**      | 62.3%          | 46.3 %     |
 
-The evaluation can be conducted using [`chainercv/examples/semantic_segmentation/eval_cityscapes.py`](https://github.com/chainer/chainercv/blob/master/examples/semantic_segmentation).
+The evaluation can be conducted using [`chainercv/examples/semantic_segmentation/eval_semantic_segmentation.py`](https://github.com/chainer/chainercv/blob/master/examples/semantic_segmentation).
 
 
 # Reference

examples/semantic_segmentation/README.md

@@ -18,15 +18,11 @@ The scores are mIoU.
 | PSPNet with ResNet101 (single scale) | 79.70 % [1] | 79.03 % |
 
 ```
-$ python eval_cityscapes.py [--model pspnet_resnet101] [--gpu <gpu>] [--pretrained-model <model_path>]
+$ python eval_semantic_segmentation.py --gpu <GPU> --dataset cityscapes --model pspnet_resnet101
+# with multiple GPUs
+$ mpiexec -n <#gpu> python eval_semantic_segmentation_multi.py --dataset cityscapes --model pspnet_resnet101
 ```
 
-You can conduct evaluation with multiple GPUs by `eval_cityscapes_multi.py`.
-Note that this script requires ChainerMN.
-
-```
-$ mpiexec -n <#gpu> python eval_cityscapes_multi.py [--model pspnet_resnet101] [--pretrained-model <model_path>]
-```
 
 ### CamVid
 
@@ -35,7 +31,7 @@ $ mpiexec -n <#gpu> python eval_cityscapes_multi.py [--model pspnet_resnet101] [
 | SegNet | 46.3 % [2] | 49.4 % |
 
 ```
-$ python eval_camvid.py [--gpu <gpu>] [--pretrained-model <model_path>] [--batchsize <batchsize>]
+$ python eval_semantic_segmentation.py --gpu <GPU> --dataset camvid --model segnet
 ```
 
 

examples/semantic_segmentation/eval_cityscapes.py → examples/semantic_segmentation/eval_semantic_segmentation.py

@@ -3,38 +3,59 @@
 import chainer
 from chainer import iterators
 
+from chainercv.datasets import ade20k_semantic_segmentation_label_names
+from chainercv.datasets import ADE20KSemanticSegmentationDataset
 from chainercv.datasets import cityscapes_semantic_segmentation_label_names
 from chainercv.datasets import CityscapesSemanticSegmentationDataset
+from chainercv.datasets import camvid_label_names
+from chainercv.datasets import CamVidDataset
+
 from chainercv.evaluations import eval_semantic_segmentation
 from chainercv.experimental.links import PSPNetResNet101
+from chainercv.links import SegNetBasic
 from chainercv.utils import apply_to_iterator
 from chainercv.utils import ProgressHook
 
 
 def main():
     parser = argparse.ArgumentParser()
     parser.add_argument(
-        '--model', choices=('pspnet_resnet101',),
-        default='pspnet_resnet101')
+        '--dataset', choices=('cityscapes', 'ade20k', 'camvid'))
+    parser.add_argument(
+        '--model', choices=(
+            'pspnet_resnet101', 'segnet'))
     parser.add_argument('--gpu', type=int, default=-1)
     parser.add_argument('--pretrained-model')
     args = parser.parse_args()
 
+    if args.dataset == 'cityscapes':
+        dataset = CityscapesSemanticSegmentationDataset(
+            split='val', label_resolution='fine')
+        label_names = cityscapes_semantic_segmentation_label_names
+    elif args.dataset == 'ade20k':
+        dataset = ADE20KSemanticSegmentationDataset(split='val')
+        label_names = ade20k_semantic_segmentation_label_names
+    elif args.dataset == 'camvid':
+        dataset = CamVidDataset(split='test')
+        label_names = camvid_label_names
+
+    if args.pretrained_model:
+        pretrained_model = args.pretrained_model
+    else:
+        pretrained_model = args.dataset
     if args.model == 'pspnet_resnet101':
-        if args.pretrained_model:
-            model = PSPNetResNet101(
-                n_class=len(cityscapes_semantic_segmentation_label_names),
-                pretrained_model=args.pretrained_model, input_size=(713, 713)
-            )
-        else:
-            model = PSPNetResNet101(pretrained_model='cityscapes')
+        model = PSPNetResNet101(
+            n_class=len(label_names),
+            pretrained_model=pretrained_model, input_size=(713, 713)
+        )
+    elif args.model == 'segnet':
+        model = SegNetBasic(
+            n_class=len(label_names), pretrained_model=pretrained_model)
 
     if args.gpu >= 0:
         chainer.cuda.get_device_from_id(args.gpu).use()
         model.to_gpu()
 
-    dataset = CityscapesSemanticSegmentationDataset(
-        split='val', label_resolution='fine')
     it = iterators.SerialIterator(
         dataset, 1, repeat=False, shuffle=False)
 
@@ -47,8 +68,7 @@ def main():
 
     result = eval_semantic_segmentation(pred_labels, gt_labels)
 
-    for iu, label_name in zip(
-            result['iou'], cityscapes_semantic_segmentation_label_names):
+    for iu, label_name in zip(result['iou'], label_names):
         print('{:>23} : {:.4f}'.format(label_name, iu))
     print('=' * 34)
     print('{:>23} : {:.4f}'.format('mean IoU', result['miou']))

examples/semantic_segmentation/eval_cityscapes_multi.py → examples/semantic_segmentation/eval_semantic_segmentation_multi.py

@@ -8,41 +8,61 @@
 
 import chainermn
 
+from chainercv.datasets import ade20k_semantic_segmentation_label_names
+from chainercv.datasets import ADE20KSemanticSegmentationDataset
 from chainercv.datasets import cityscapes_semantic_segmentation_label_names
 from chainercv.datasets import CityscapesSemanticSegmentationDataset
+from chainercv.datasets import camvid_label_names
+from chainercv.datasets import CamVidDataset
+
 from chainercv.evaluations import calc_semantic_segmentation_confusion
 from chainercv.evaluations import calc_semantic_segmentation_iou
 from chainercv.experimental.links import PSPNetResNet101
+from chainercv.links import SegNetBasic
 from chainercv.utils import apply_to_iterator
 from chainercv.utils import ProgressHook
 
 
 def main():
     parser = argparse.ArgumentParser()
     parser.add_argument(
-        '--model', choices=('pspnet_resnet101',),
-        default='pspnet_resnet101')
+        '--dataset', choices=('cityscapes', 'ade20k', 'camvid'))
+    parser.add_argument(
+        '--model', choices=(
+            'pspnet_resnet101', 'segnet'))
     parser.add_argument('--pretrained-model')
     args = parser.parse_args()
 
     comm = chainermn.create_communicator()
     device = comm.intra_rank
 
+    if args.dataset == 'cityscapes':
+        dataset = CityscapesSemanticSegmentationDataset(
+            split='val', label_resolution='fine')
+        label_names = cityscapes_semantic_segmentation_label_names
+    elif args.dataset == 'ade20k':
+        dataset = ADE20KSemanticSegmentationDataset(split='val')
+        label_names = ade20k_semantic_segmentation_label_names
+    elif args.dataset == 'camvid':
+        dataset = CamVidDataset(split='test')
+        label_names = camvid_label_names
+
+    if args.pretrained_model:
+        pretrained_model = args.pretrained_model
+    else:
+        pretrained_model = args.dataset
     if args.model == 'pspnet_resnet101':
-        if args.pretrained_model:
-            model = PSPNetResNet101(
-                n_class=len(cityscapes_semantic_segmentation_label_names),
-                pretrained_model=args.pretrained_model, input_size=(713, 713)
-            )
-        else:
-            model = PSPNetResNet101(pretrained_model='cityscapes')
+        model = PSPNetResNet101(
+            n_class=len(label_names),
+            pretrained_model=pretrained_model, input_size=(713, 713)
+        )
+    elif args.model == 'segnet':
+        model = SegNetBasic(
+            n_class=len(label_names), pretrained_model=pretrained_model)
 
     chainer.cuda.get_device_from_id(device).use()
     model.to_gpu()
 
-    dataset = CityscapesSemanticSegmentationDataset(
-        split='val', label_resolution='fine')
-
     if comm.rank == 0:
         indices = np.arange(len(dataset))
     else:
@@ -68,8 +88,7 @@ def main():
         pixel_accuracy = np.diag(confusion).sum() / confusion.sum()
         class_accuracy = np.diag(confusion) / np.sum(confusion, axis=1)
 
-        for iu, label_name in zip(
-                iou, cityscapes_semantic_segmentation_label_names):
+        for iu, label_name in zip(iou, label_names):
             print('{:>23} : {:.4f}'.format(label_name, iu))
         print('=' * 34)
         print('{:>23} : {:.4f}'.format('mean IoU', np.nanmean(iou)))

examples_tests/semantic_segmentation_tests/test_eval_semantic_segmentation.sh

@@ -0,0 +1,9 @@
+cd examples/semantic_segmentation
+
+sed -e 's/CamVidDataset(split='\''test'\'')/CamVidDataset(split='\''test'\'').slice[:20]/' -i eval_camvid.py
+$PYTHON eval_semantic_segmentation.py --gpu 0 --dataset camvid --model segnet
+
+sed -e 's/label_resolution='\''fine'\'')/label_resolution='\''fine'\'').slice[:20]/' \
+    -i eval_cityscapes.py
+$PYTHON eval_semantic_segmentation.py --gpu 0 --dataset cityscapes --model pspnet_resnet101
+

examples_tests/semantic_segmentation_tests/test_eval_semantic_segmentation_multi.sh

@@ -0,0 +1,8 @@
+cd examples/semantic_segmentation
+
+sed -e 's/CamVidDataset(split='\''test'\'')/CamVidDataset(split='\''test'\'').slice[:20]/' -i eval_camvid.py
+$MPIEXEC $PYTHON eval_semantic_segmentation_multi.py --gpu 0 --dataset camvid --model segnet
+
+sed -e 's/label_resolution='\''fine'\'')/label_resolution='\''fine'\'').slice[:20]/' \
+    -i eval_cityscapes.py
+$MPIEXEC $PYTHON eval_semantic_segmentation_multi.py --gpu 0 --dataset cityscapes --model pspnet_resnet101