Gait-PowerShot-SX430-IS-A, background and Filtered Silhouette Images

Gait-PowerShot-SX430-IS-A Dataset

It is a gait dataset of 130 individuals, recorded indoors under controlled night-time illumination with a Canon PowerShot SX430 IS camera mounted on an approximately fixed tripod in a fronto-parallel setup. Each subject walked naturally across the camera’s field of view; after segmentation this yields ≈520 sequences (about four sequences extracted per subject). Silhouette images were produced by background subtraction followed by morphological processing (threshold = 56, kernel size = 4) and normalization to a fixed image size. The dataset is intended for research in silhouette-based and appearance-based gait recognition, and used to construct the Aggregated Pose Energy Image (APEI) feature.

What’s included in this repository:

Silhouette images for all subjects. The silhouettes are the inputs used to build the APEI feature.

Code:

The first three Python scripts required to extract the silhouettes images are available on GitHub:
https://github.com/rspvsanjay/Aggregated-Pose-Energy-Image/

Notes & access:

The original raw video recordings are not included in this repository but can be provided upon reasonable request.

For access to the videos or other inquiries, contact: [e-mail]