Research projects of person re-identification
The purpose of person re-identification is to recognize a unique pedestrian among different camera views.
The pedestrian should have little discrepancy in appearance, and the cameras should be still.
A typical input is two sets of pedestrians images sequences, namely gallery and probe. The output should be a distance matrix. Cumulative Match Characteristic (CMC) curve is used for evaluation.
At current stage of research, no setup scripts are provided. Please just export this project's path to PYTHONPATH.