Founded in Singapore in 2015, KLASS was established with the goal of advancing applied Research & Development work as well as bespoke solution creation in the local public safety space. We recognise the importance of innovation in bringing cutting-edge technologies to augment the current public safety offerings and meet the growing risks associated with the evolving digital, socio-economic and environmental landscapes. Through championing the adoption of emerging technologies in public safety, we hope to enhance operational capabilities to deliver targeted mission outcomes.
Our continued investment in innovation is two-pronged, focusing on both originating innovation internally and supporting external innovation projects with high potential to address client problem statements.
One such internally originated project is GROWN+UP, a one-size-fits-all webpage parser which can extract useful webpage features. GROWN+Up can outperform other methods of webpage information retrieval by fine-tuning on small datasets.
We have also supported other exciting innovations such as the use of Viseme-Phoneme Mapping and Transfer for Audio-Visual speech recognition. By using viseme, which represents the position of the face and mouth when saying a word, viseme and sound features are first extracted and mapped to corresponding clusters during training via proposed online balanced clustering. Further, an Adversarial Mutual Information Estimator is used to maximize the mutual information between clustered visemes and phonemes to learn viseme-phoneme mapping. This helps to overcome the ambiguity of homophones and using this learnt mapping, modality transfer is implemented to restore clean audio from lips whenever SNR (signal to noise ratio) of speech is poor. Experimental results in public benchmarks showed that Uni-VPM achieves state-of-the-art under various noisy as well as clean conditions. Additionally, this approach can outperform existing techniques on visual speech recognition. This work is achieved through a KLASS research collaboration with Nanyang Technological University, supported by National Research Foundation, Singapore under its AI Singapore Programme.
Fig. 1: Discussion with Assoc Prof Chng’s team
Reflecting on the collaborative partnership, Assoc Prof. Chng Eng Siong from NTU AISG Speech Lab remarked, “Our collaboration with KLASS spans over 5 fruitful years, resulting in the technological transfer of speech recognition technologies and diarisation. We anticipate more fruitful years ahead with KLASS!”
Among our supported projects that garnered significant public attention and interest is that of our Cyborg Insect Project, which was featured at the recently concluded MAP-TXS 24 event.
Fig. 2: Cyborg Insect Exhibit at the HTX booth (MAP-TXS 24)
The project currently entails the mounting of a miniature backpack (15g) of sensors, comprising infrared camera and processor, onto a Madagascar Hissing cockroach (Gromphadorhina portentosa). Presently, the insects can be remotely steered or configured to autonomously navigate their environment, with future development aiming to refine the tracking and communication systems to be used in search and rescue operations, locating injured or trapped individuals.
Our collaboration with Professor Hirotaka Sato and his team from Nanyang Technological University (NTU) has led to the practical implementation of control, localisation, and imaging utilising cyborg insects in real-world scenarios. Alex Low, Senior Software Engineer (AIoT), who collaborated closely with the NTU team on this project, expressed his optimism for a future in which “these assets are deployed globally for search and rescue missions, enabling responders to take advantage of the critical initial moments. The adoption of such technology is expected to enhance rescuer safety and mitigate risks during operations.”
Looking ahead, the team aims to incorporate a broader range of sensors (such as RGB cameras and mmWave presence sensors) and state-of-the-art technologies to extend the application of these cyborg insects beyond search and rescue missions.
We continuously seek out cutting-edge technologies that can elevate public safety in Singapore. If you possess an innovative idea or solution and wish to collaborate with us to realise its potential, please reach out to us at marketing@klasses.com.sg.
