Overview of Open-ReID

Open Re-ID is a lightweight library of person re-identification for research purpose. It aims to provide a uniform interface for different datasets, a full set of models and evaluation metrics, as well as examples to reproduce (near) state-of-the-art results. Open-ReID is mainly based on PyTorch.

Structure

Open-ReID is structured into three levels, as shown in the figure below.

API Level

At bottom, there are decoupled modules each providing unit functions. For example, the datasets module has a uniform interface for many popular datasets, while commonly used evaluation metrics, such as CMC and mean AP are implemented in the evaluation_metrics module, which accept both torch.Tensor and numpy.ndarray as inputs.

SDK Level

In the middle, several classes interact with underlying APIs to provide routines for standard tasks. For example, the Trainer can be used to train a deep model on training set, and Evaluator can evaluate the model on validation and test sets.

Application Level

At top, we provide several examples using Open-ReID. For example, one can easily train a CNN with different kinds of loss functions on different datasets, to achieve certain baselines or state-of-the-art results.