Skip to content

Latest commit

 

History

History
61 lines (44 loc) · 2.22 KB

preprocessing.md

File metadata and controls

61 lines (44 loc) · 2.22 KB
Raw

Instructions to precompute data required for training

Compiling SUNCG Toolbox

cd SUNCG_DIR;

# Download the toolbox
git clone https://github.com/shurans/SUNCGtoolbox ./toolbox
cd toolbox

# Use our modified rendering function
cp CODE_ROOT/preprocess/suncg/scn2img.cpp ./gaps/apps/scn2img/

# Compile
make

# (or optionally instead of above) compile with offscreen mesa support
make mesa

In case you compile with offscreen support, you might need to edit this line to specify additional lib directories if you're using a locally compiled version of mesa, and possibly also specify a CPLUS_INCLUDE_PATH. Though note that locally compiling mesa with offscreen support can get a bit tricky.

We highly recommend using the offscreen version, as otherwise the rendering behaviour is often stochastic.

Rendering Layout and Node Images

You'll first need to edit the 'sunCgDir' variable in both the python scripts below. Note that both the rendering jobs can take a while. If you managed to compile the gaps toolbox with offscreen mesa, you can add --mesa=True to the commands below, else you'll need to run the rendering jobs in an onscreen mode.

cd CODE_ROOT/preprocess/suncg

# Render amodal depths (edit the 'sunCgDir' variable before running)
python render_layout_depth.py --min=1 --nc=1

# Render node indices (edit the 'sunCgDir' variable before running)
python render_node_indices.py --min=1 --nc=1

Voxelize Objects and Scenes

Please download binvox from here and store the binary as SUNCG_DIR/toolbox/binvox.

# Voxelize the objects (edit the 'sunCgDir' variable before running)
# This needs to be run in onscreen mode with a display/desktop connected
python voxelize_objects.py

# Compute voxelizations for the full scene (required for training the baseline)
# Edit the 'suncgDir' in globals.m before running
precompute_scene_voxels(1, 0);

Compute object proposals

# Extract ground-truth object boxes
# Edit the 'suncgDir' in globals.m before running
precompute_gt_bboxes(1, 0);

# Extract edgebox proposals
# Edit the 'suncgDir' in globals.m before running
precompute_edge_boxes(1, 0);