This tutorial will help you, step-by-step, how to extract features (Angles and Trajectories) from 2D Poses.
Before proceeding make sure that you have already extract or download 2D poses, see 2D Poses Extraction for more information.
Before proceeding make sure that you have already encode 2D poses in Parameter Space, see Encoding 2D poses in Space Param for more information.
Experiments were performed in two public dataset KTH and Weizmann.
In this example, we will extract features from 2D poses by running the following scripts.
KTH dataset example:
python tools/FeaturesExtraction/WEF/compute_angles_from_body_parts.py \
--poses_base_dir=/home/murilo/dataset/KTH \
--input_dir=2DPoses_Person \
--output_dir=Angles_from_2DPoses
Weizmann dataset example:
python tools/FeaturesExtraction/WEF/compute_angles_from_body_parts.py \
--poses_base_dir=/home/murilo/dataset/Weizmann \
--input_dir=2DPoses \
--output_dir=Angles_from_2DPoses
KTH dataset example:
python tools/FeaturesExtraction/PEF/compute_trajectories_from_parameter_space.py \
--space_param_base_dir=/home/murilo/dataset/KTH \
--input_dir=2DPoses_SpaceParam \
--output_dir=2DPoses_SpaceParam_Trajectories \
--number_frames=20 --stride=10
Weizmann dataset example:
python tools/FeaturesExtraction/PEF/compute_trajectories_from_parameter_space.py \
--space_param_base_dir=/home/murilo/dataset/Weizmann \
--input_dir=2DPoses_SpaceParam \
--output_dir=2DPoses_SpaceParam_Trajectories \
--number_frames=20 --stride=1
As next step follow the link: Human Action Recognition