Useful tools
extract images from videos: apps/preprocess/extract_image.py
python3 apps/preprocess/extract_image.py ${data}
Extra flags:
flag | type | |
---|---|---|
–image | str | output directory(default images ) |
–ffmpeg | str | path to ffmpeg(default ffmpeg ) |
–transpose | int | transpose for ffmepeg(default -1, for clockwise) |
copy and sample dataset: scripts/preprocess/copy_dataset.py
python3 scripts/preprocess/copy_dataset.py /path/to/input /path/to/output
Extra flags:
flag | type | |
---|---|---|
–start | int | start frame(default 0) |
–end | int | end frame(default None) |
–step | int | sample step(default 1) |
–scale | float | resize image with scale(default 1) |
Export vertices
python3 apps/postprocess/write_vertices.py ${data}/output-vposer-3d/smpl ${data}/output-vposer-3d/vertices --cfg_model ${data}/output-vposer-3d/cfg_model.yml --mode vertices
Export triangle mesh
python3 apps/postprocess/write_vertices.py ${data}/output-vposer-3d/smpl ${data}/output-vposer-3d/mesh --cfg_model ${data}/output-vposer-3d/cfg_model.yml --mode mesh
MoCap ==> NeuralBody
# convert to vertices
python3 apps/postprocess/write_vertices.py ${data}/output-smpl-3d/smpl ${data}/output-smpl-3d/vertices --cfg_model ${data}/output-smpl-3d/cfg_model.yml --mode vertices
# render mask
python3 apps/postprocess/render.py ${data} --exp output-smpl-3d --mode instance-d0.05 --ranges 0 1400 1