A true interpreting agent not only understands sign language and translates to text, but also understands text and translates to signs. Much of the AI work in sign language translation to date has focused mainly on translating from signs to text. Towards the latter goal, we propose a text-to-sign translation model, SignNet, which exploits the notion of similarity (and dissimilarity) of visual signs in translating. This module presented is only one part of a dual-learning two task process involving text-to-sign (T2S) as well as sign-to-text (S2T). We currently implement SignNet as a single channel architecture so that the output of the T2S task can be fed into S2T in a continuous dual learning framework. By single channel, we refer to a single modality, the body pose joints. In this work, we present SignNet, a T2S task using a novel metric embedding learning process, to preserve the distances between sign embeddings relative to their dissimilarity. We also describe how to choose positive and negative examples of signs for similarity testing. From our analysis, we observe that metric embedding learning-based model perform significantly better than the other models with traditional losses, when evaluated using BLEU scores. In the task of gloss to pose, SignNet performed as well as its state-of-the-art (SoTA) counterparts and outperformed them in the task of text to pose, by showing noteworthy enhancements in BLEU 1 - BLEU 4 scores (BLEU 1: 31 → 39; ≈26% improvement and BLEU 4: 10.43 →11.84; ≈14% improvement) when tested on the popular RWTH PHOENIX-Weather-2014T benchmark dataset
more »
« less
SignNet II: A Transformer-Based Two-Way Sign Language Translation Model
The role of a sign interpreting agent is to bridge the communication gap between the hearing-only and Deaf or Hard of Hearing communities by translating both from sign language to text and from text to sign language. Until now, much of the AI work in automated sign language processing has focused primarily on sign language to text translation, which puts the advantage mainly on the side of hearing individuals. In this work, we describe advances in sign language processing based on transformer networks. Specifically, we introduce SignNet II, a sign language processing architecture, a promising step towards facilitating two-way sign language communication. It is comprised of sign-to-text and text-to-sign networks jointly trained using a dual learning mechanism. Furthermore, by exploiting the notion of sign similarity, a metric embedding learning process is introduced to enhance the text-to-sign translation performance. Using a bank of multi-feature transformers, we analyzed several input feature representations and discovered that keypoint-based pose features consistently performed well, irrespective of the quality of the input videos. We demonstrated that the two jointly trained networks outperformed their singly-trained counterparts, showing noteworthy enhancements in BLEU-1 - BLEU-4 scores when tested on the largest available German Sign Language (GSL) benchmark dataset.
more »
« less
- Award ID(s):
- 2223507
- NSF-PAR ID:
- 10395319
- Date Published:
- Journal Name:
- IEEE Transactions on Pattern Analysis and Machine Intelligence
- ISSN:
- 0162-8828
- Page Range / eLocation ID:
- 1 to 14
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Sign language translation without transcription has only recently started to gain attention. In our work, we focus on improving the state-of-the-art translation by introducing a multi-feature fusion architecture with enhanced input features. As sign language is challenging to segment, we obtain the input features by extracting overlapping scaled segments across the video and obtaining their 3D CNN representations. We exploit the attention mechanism in the fusion architecture by initially learning dependencies between different frames of the same video and later fusing them to learn the relations between different features from the same video. In addition to 3D CNN features, we also analyze pose-based features. Our robust methodology outperforms the state-of-the-art sign language translation model by achieving higher BLEU 3 – BLEU 4 scores and also outperforms the state-of-the-art sequence attention models by achieving a 43.54% increase in BLEU 4 score. We conclude that the combined effects of feature scaling and feature fusion make our model more robust in predicting longer n-grams which are crucial in continuous sign language translation.more » « less
-
While a significant amount of work has been done on the commonly used, tightly -constrained weather-based, German sign language (GSL) dataset, little has been done for continuous sign language translation (SLT) in more realistic settings, including American sign language (ASL) translation. Also, while CNN - based features have been consistently shown to work well on the GSL dataset, it is not clear whether such features will work as well in more realistic settings when there are more heterogeneous signers in non-uniform backgrounds. To this end, in this work, we introduce a new, realistic phrase-level ASL dataset (ASLing), and explore the role of different types of visual features (CNN embeddings, human body keypoints, and optical flow vectors) in translating it to spoken American English. We propose a novel Transformer-based, visual feature learning method for ASL translation. We demonstrate the explainability efficacy of our proposed learning methods by visualizing activation weights under various input conditions and discover that the body keypoints are consistently the most reliable set of input features. Using our model, we successfully transfer-learn from the larger GSL dataset to ASLing, resulting in significant BLEU score improvements. In summary, this work goes a long way in bringing together the AI resources required for automated ASL translation in unconstrained environments.more » « less
-
null (Ed.)Deaf spaces are unique indoor environments designed to optimize visual communication and Deaf cultural expression. However, much of the technological research geared towards the deaf involve use of video or wearables for American sign language (ASL) translation, with little consideration for Deaf perspective on privacy and usability of the technology. In contrast to video, RF sensors offer the avenue for ambient ASL recognition while also preserving privacy for Deaf signers. Methods: This paper investigates the RF transmit waveform parameters required for effective measurement of ASL signs and their effect on word-level classification accuracy attained with transfer learning and convolutional autoencoders (CAE). A multi-frequency fusion network is proposed to exploit data from all sensors in an RF sensor network and improve the recognition accuracy of fluent ASL signing. Results: For fluent signers, CAEs yield a 20-sign classification accuracy of %76 at 77 GHz and %73 at 24 GHz, while at X-band (10 Ghz) accuracy drops to 67%. For hearing imitation signers, signs are more separable, resulting in a 96% accuracy with CAEs. Further, fluent ASL recognition accuracy is significantly increased with use of the multi-frequency fusion network, which boosts the 20-sign fluent ASL recognition accuracy to 95%, surpassing conventional feature level fusion by 12%. Implications: Signing involves finer spatiotemporal dynamics than typical hand gestures, and thus requires interrogation with a transmit waveform that has a rapid succession of pulses and high bandwidth. Millimeter wave RF frequencies also yield greater accuracy due to the increased Doppler spread of the radar backscatter. Comparative analysis of articulation dynamics also shows that imitation signing is not representative of fluent signing, and not effective in pre-training networks for fluent ASL classification. Deep neural networks employing multi-frequency fusion capture both shared, as well as sensor-specific features and thus offer significant performance gains in comparison to using a single sensor or feature-level fusion.more » « less
-
Many sign languages are bona fide natural languages with grammatical rules and lexicons hence can benefit from machine translation methods. Similarly, since sign language is a visual-spatial language, it can also benefit from computer vision methods for encoding it. With the advent of deep learning methods in recent years, significant advances have been made in natural language processing (specifically neural machine translation) and in computer vision methods (specifically image and video captioning). Researchers have therefore begun expanding these learning methods to sign language understanding. Sign language interpretation is especially challenging, because it involves a continuous visual-spatial modality where meaning is often derived based on context. The focus of this article, therefore, is to examine various deep learning–based methods for encoding sign language as inputs, and to analyze the efficacy of several machine translation methods, over three different sign language datasets. The goal is to determine which combinations are sufficiently robust for sign language translation without any gloss-based information. To understand the role of the different input features, we perform ablation studies over the model architectures (input features + neural translation models) for improved continuous sign language translation. These input features include body and finger joints, facial points, as well as vector representations/embeddings from convolutional neural networks. The machine translation models explored include several baseline sequence-to-sequence approaches, more complex and challenging networks using attention, reinforcement learning, and the transformer model. We implement the translation methods over multiple sign languages—German (GSL), American (ASL), and Chinese sign languages (CSL). From our analysis, the transformer model combined with input embeddings from ResNet50 or pose-based landmark features outperformed all the other sequence-to-sequence models by achieving higher BLEU2-BLEU4 scores when applied to the controlled and constrained GSL benchmark dataset. These combinations also showed significant promise on the other less controlled ASL and CSL datasets.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.1109/TPAMI.2022.3232389
