This archive contains the RWTH-Weather-Phoenix 2014 signer independent SI5 continuous sign language recognition corpus.
It is released under non-commercial cc 4.0 license with attribution (see attachment)

If you use this data in your research, please cite:

O. Koller, J. Forster, and H. Ney. Continuous sign language recognition: Towards large vocabulary statistical recognition systems handling multiple signers. Computer Vision and Image Understanding, volume 141, pages 108-125, December 2015.

(for the general corpus)

and

Koller, Zargaran, Ney. "Re-Sign: Re-Aligned End-to-End Sequence Modeling with Deep Recurrent CNN-HMMs" in CVPR 2017, Honululu, Hawaii, USA.

(for the signer independent setup)


The signer independent SI5 partition contains 8 signers in train and 1 signer (unseen signer, which is signer 5) in test and has been recorded on the broadcastnews channel.

phoenix-2014-signerindependent-SI5
├── annotations 
│   │
│   └── manual -> this contains the corpus files
│
├── evaluation -> this contains an evaluation script. make sure to have a compiled version of the NIST sclite tools in your path. call: ./evaluatePhoenix2014.sh example-hypothesis-dev.ctm dev
│
├── features
│    │
│    └── fullFrame-210x260px -> resolution of 210x260 pixels, but they are distorted due to transmission channel particularities, to undistort stretch images to 210x300
│        ├── dev
│        ├── test
│        └── train
│
│
└── models -> we provide caffe models to achieve the published  45.1 / 44.1 % WER on the dev/test partition of this corpus, we also provide our languagemodel