Experiments on MIT Data

Experimental Question

From Massachusetts Institute of Technology's database, the performance of our gait recognition methodology on video data (of walking subjects) collected over a number of days spanning 3 months is evaluated. Their database contains only frontal-parallel views for gait recognition.

For a complete description of their database see their description page .

Experiments and Results

We generate tables that are similarity comparison (by finding the L2norm) between gallery and probe views for the following experiments. The similarity matrixes are presented as text files and cmc curves. The headings of the columns refer to the gallery, and the row headings are the probe view(s). The file labels for the headings can be found on MIT's evaluation page. For out technique there is not a training set.

Experiment 2: Comparing sequences from one day against sequences on other days. This experiment explores the robustness of gait measurements across different days.

Experiment Specifics

Our paradigm of measuring static body parameters does not require any specific training set that helps define a model that is then applied to the gallery and the probes.  When performing matching across views, all subjects are used to compute "cross-condition mapping functions".  That is, a single linear regression was done for each measured feature in our method that would map from one condition to the other.  All subjects were used because of the limited number of subjects. Also, we were unable to use our depth compensation method because of a lack of calibration information.



Copyright 1997-2001