Skip to content

Commit

Permalink
PASCAL VOC: include more data details, rename layers -> voc_layers
Browse files Browse the repository at this point in the history
  • Loading branch information
shelhamer committed May 19, 2016
1 parent 07e74ff commit 9661609
Show file tree
Hide file tree
Showing 7 changed files with 48 additions and 3 deletions.
21 changes: 21 additions & 0 deletions data/pascal/README.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,21 @@
# PASCAL VOC and SBD

PASCAL VOC is a standard recognition dataset and benchmark with detection and semantic segmentation challenges.
The Semantic Boundary Dataset (SBD) is a further annotation of the PASCAL VOC data that provides more semantic segmentation and instance segmentation masks.

PASCAL VOC has a private test set and [leaderboard for semantic segmentation](http://host.robots.ox.ac.uk:8080/leaderboard/displaylb.php?challengeid=11&compid=6).

The train/val/test splits of PASCAL VOC segmentation challenge and SBD diverge.
Most notably VOC 2011 segval intersects with SBD train.
Care must be taken for proper evaluation by excluding images from the train or val splits.

We train on the 8,498 images of SBD train.
We validate on the non-intersecting set defined in the included `seg11valid.txt`.

Refer to `classes.txt` for the listing of classes in model output order.
Refer to `../voc_layers.py` for the Python data layer for this dataset.

See the dataset sites for download:

- PASCAL VOC 2012: http://host.robots.ox.ac.uk/pascal/VOC/voc2012/
- SBD: http://www.cs.berkeley.edu/~bharath2/codes/SBD/download.html
24 changes: 24 additions & 0 deletions data/pascal/classes.txt
Original file line number Diff line number Diff line change
@@ -0,0 +1,24 @@
background
aeroplane
bicycle
bird
boat
bottle
bus
car
cat
chair
cow
diningtable
dog
horse
motorbike
person
pottedplant
sheep
sofa
train
tvmonitor

and 255 is the ignore label that marks pixels excluded from learning and
evaluation by the PASCAL VOC ground truth.
File renamed without changes.
2 changes: 1 addition & 1 deletion voc-fcn16s/net.py
Original file line number Diff line number Diff line change
Expand Up @@ -21,7 +21,7 @@ def fcn(split):
else:
pydata_params['voc_dir'] = '../../data/pascal/VOC2011'
pylayer = 'VOCSegDataLayer'
n.data, n.label = L.Python(module='layers', layer=pylayer,
n.data, n.label = L.Python(module='voc_layers', layer=pylayer,
ntop=2, param_str=str(pydata_params))

# the base net
Expand Down
2 changes: 1 addition & 1 deletion voc-fcn32s/net.py
Original file line number Diff line number Diff line change
Expand Up @@ -21,7 +21,7 @@ def fcn(split):
else:
pydata_params['voc_dir'] = '../../data/pascal/VOC2011'
pylayer = 'VOCSegDataLayer'
n.data, n.label = L.Python(module='layers', layer=pylayer,
n.data, n.label = L.Python(module='voc_layers', layer=pylayer,
ntop=2, param_str=str(pydata_params))

# the base net
Expand Down
2 changes: 1 addition & 1 deletion voc-fcn8s/net.py
Original file line number Diff line number Diff line change
Expand Up @@ -21,7 +21,7 @@ def fcn(split):
else:
pydata_params['voc_dir'] = '../../data/pascal/VOC2011'
pylayer = 'VOCSegDataLayer'
n.data, n.label = L.Python(module='layers', layer=pylayer,
n.data, n.label = L.Python(module='voc_layers', layer=pylayer,
ntop=2, param_str=str(pydata_params))

# the base net
Expand Down
File renamed without changes.

0 comments on commit 9661609

Please sign in to comment.