Hidden Conditional Random Fields for Visual Speech Recognition

In this paper we present the application of Hidden Conditional Random Fields (HCRFs) to modelling speech for visual speech recognition. HCRFs may be easily adapted to model long range dependencies across an observation sequence. As a result visual word recognition performance can be improved as the model is able to take more of a contextual approach to generating state sequences. Results are presented from a speaker-dependent, isolated digit, visual speech recognition task using comparisons with a baseline HMM system. We firstly illustrate that word recognition rates on clean video using HCRFs can be improved by increasing the number of past and future observations being taken into account by each state. Secondly we compare model performances using various levels of video compression on the test set. As far as we are aware this is the first attempted use of HCRFs for visual speech recognition.

Inter-Frame Contextual Modelling For Visual Speech Recognition

In this paper, we present a new approach to visual speech recognition which improves contextual modelling by combining Inter-Frame Dependent and Hidden Markov Models. This approach captures contextual information in visual speech that may be lost using a Hidden Markov Model alone. We apply contextual modelling to a large speaker independent isolated digit recognition task, and compare our approach to two commonly adopted feature based techniques for incorporating speech dynamics. Results are presented from baseline feature based systems and the combined modelling technique. We illustrate that both of these techniques achieve similar levels of performance when used independently. However significant improvements in performance can be achieved through a combination of the two. In particular we report an improvement in excess of 17% relative Word Error Rate in comparison to our best baseline system.

Fast recognition of musical sounds based on timbre

Human listeners seem to have an impressive ability to recognize a wide variety of natural sounds. However, there is surprisingly little quantitative evidence to characterize this fundamental ability. Here the speed and accuracy of musical-sound recognition were measured psychophysically with a rich but acoustically balanced stimulus set. The set comprised recordings of notes from musical instruments and sung vowels. In a first experiment, reaction times were collected for three target categories: voice, percussion, and strings. In a go/no-go task, listeners reacted as quickly as possible to members of a target category while withholding responses to distractors (a diverse set of musical instruments). Results showed near-perfect accuracy and fast reaction times, particularly for voices. In a second experiment, voices were recognized among strings and vice-versa. Again, reaction times to voices were faster. In a third experiment, auditory chimeras were created to retain only spectral or temporal features of the voice. Chimeras were recognized accurately, but not as quickly as natural voices. Altogether, the data suggest rapid and accurate neural mechanisms for musical-sound recognition based on selectivity to complex spectro-temporal signatures of sound sources.

Episodic Reasoning for Vision-Based Human Action Recognition

Smart Spaces, Ambient Intelligence, and Ambient Assisted Living are environmental paradigms that strongly depend on their capability to recognize human actions. While most solutions rest on sensor value interpretations and video analysis applications, few have realized the importance of incorporating common-sense capabilities to support the recognition process. Unfortunately, human action recognition cannot be successfully accomplished by only analyzing body postures. On the contrary, this task should be supported by profound knowledge of human agency nature and its tight connection to the reasons and motivations that explain it. The combination of this knowledge and the knowledge about how the world works is essential for recognizing and understanding human actions without committing common-senseless mistakes. This work demonstrates the impact that episodic reasoning has in improving the accuracy of a computer vision system for human action recognition. This work also presents formalization, implementation, and evaluation details of the knowledge model that supports the episodic reasoning.

Auditory gist: Recognition of very short sounds from timbre cues

Sounds such as the voice or musical instruments can be recognized on the basis of timbre alone. Here, sound recognition was investigated with severely reduced timbre cues. Short snippets of naturally recorded sounds were extracted from a large corpus. Listeners were asked to report a target category (e.g., sung voices) among other sounds (e.g., musical instruments). All sound categories covered the same pitch range, so the task had to be solved on timbre cues alone. The minimum duration for which performance was above chance was found to be short, on the order of a few milliseconds, with the best performance for voice targets. Performance was independent of pitch and was maintained when stimuli contained less than a full waveform cycle. Recognition was not generally better when the sound snippets were time-aligned with the sound onset compared to when they were extracted with a random starting time. Finally, performance did not depend on feedback or training, suggesting that the cues used by listeners in the artificial gating task were similar to those relevant for longer, more familiar sounds. The results show that timbre cues for sound recognition are available at a variety of time scales, including very short ones.

Face Recognition in the Scrambled Domain via Salience-Aware Ensembles of Many Kernels

With the rapid development of internet-of-things (IoT), face scrambling has been proposed for privacy protection during IoT-targeted image/video distribution. Consequently in these IoT applications, biometric verification needs to be carried out in the scrambled domain, presenting significant challenges in face recognition. Since face models become chaotic signals after scrambling/encryption, a typical solution is to utilize traditional data-driven face recognition algorithms. While chaotic pattern recognition is still a challenging task, in this paper we propose a new ensemble approach – Many-Kernel Random Discriminant Analysis (MK-RDA) to discover discriminative patterns from chaotic signals. We also incorporate a salience-aware strategy into the proposed ensemble method to handle chaotic facial patterns in the scrambled domain, where random selections of features are made on semantic components via salience modelling. In our experiments, the proposed MK-RDA was tested rigorously on three human face datasets: the ORL face dataset, the PIE face dataset and the PUBFIG wild face dataset. The experimental results successfully demonstrate that the proposed scheme can effectively handle chaotic signals and significantly improve the recognition accuracy, making our method a promising candidate for secure biometric verification in emerging IoT applications.