Skip to content

Latest commit

 

History

History
81 lines (66 loc) · 3.52 KB

README.md

File metadata and controls

81 lines (66 loc) · 3.52 KB
Raw

parse27k_tools

Tools for the Parse-27k Dataset - evaluation routines and some simple scripts to get started...

Find the dataset here: [http://www.vision.rwth-aachen.de/parse27k]

Installation & Requirements

We use Python 2.7 on a 64-bit Linux -- but the tools should run on other platforms as well.

The tools have some dependencies. If you are using Python regularly for scientific computing, all of this is likely installed already. Otherwise they can all be easily obtained through pip. These tools depend (directly or indirectly) on:

futures
progressbar2
h5py
Pillow
scipy
scikit-image
matplotlib

Again, if you have a working Python working environment, most of these will already be installed.

Using this in case you have no Python Environment

If you have never used Python before, install Python 2.7 on your system through your favorite installation method (e.g. apt-get). I would then recommend to setup a virtualenv with the dependencies. For this, follow these commands:

virtualenv new_environment
source new_environment/bin/activate
pip install numpy
pip install scipy
pip install scikit-image
pip install -r requirements.txt

Note due to some build problems, which I only partially understand, I could not directly install all requirements with a single pip call. It did work, to first install numpy,scipy, and scikit-image separately. If anybody can point me to a nicer solution, I will be happy to adapt this.

Usage

You can run the scripts with option -h for a basic description of their parameters. The basic functions are:

  • preprocess_dataset.py: create crops from the full images and store in HDF5 file.
  • parse_evaluation.py: compute performance measures from predictions.
  • visualize_crops.py: visualize examples from a file generated by preprocess_dataset.py.
  • visualize_pid.py: visualize a specific pedestrianID from an HDF5 file generated by preprocess_dataset.py. The pedestrianID is a unique identifier for the examples within the dataset.

Preprocessing the Dataset

As an example of how to use the dataset, this package includes preprocess_dataset.py. This will read the annotations from the database, crop the examples from the original images, and store the result in an HDF5 file. The resulting file will have two datasets (as it is called in HDF5 language): crops and labels. The labels is a NxK matrix, for N examples and K attributes. The meaning is encoded in the names attribute.

$ export PARSE_PATH=_where_you_saved_the_dataset_
$ ./preprocess_dataset.py --output_dir /tmp/crops --padding_mode zero --padding 32
$ ./visualize_crops.py -r /tmp/crops/train.hdf5

Evaluation Modes

The tool parse_evaluation.py allows two modes of evaluation - retrieval and classification. See our paper for details on the difference.

The tool expects the predictions in an HDF5-file, with one dataset for each attribute. You can use the make_up_predictions.py script to learn about the expected format.

$ ./parse_evaluation.py /tmp/crops/test.hdf5 ./random_predictions.hdf5

Bugs & Problems & Updates

The tools here were taken from a larger set of tools as we prepared the dataset for publication. In case we notice problems or bugs introduced in this process, we will fix these here.

If you are working with the Parse-27k dataset, we encourage you to follow this repository. In case there are any bug-fixes or changes to the evaluation pipeline, GitHub will let you know.