Event-based diffractive neural network chip for dynamic action recognition

Li, Zeying; Su, Hang; Li, Baoli; Luan, Haitao; Gu, Min; Fang, Xinyuan

doi:10.1016/j.optlastec.2023.110136

Cited by 7 publications

(2 citation statements)

References 53 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These advancements facilitate visualization through random diffusers and enable the extraction of phase details from objects positioned behind such diffusers. A latest study has unveiled the creation of an event-based DNN chip specifically designed for dynamic human action recognition, leveraging an event camera and achieving an impressive accuracy rate of approximately 99% [202]. In addition, optical DNNs can be used to achieve the multiplexing and demultiplexing functions, which is important to further increase the capacity of information computing [203,204].…”

Section: Computing Imaging With Diffractive Optical Neural Networkmentioning

confidence: 99%

Two-photon polymerization lithography for imaging optics

Wang,

Pan,

et al. 2024

Int. J. Extrem. Manuf.

View full text Add to dashboard Cite

Optical imaging systems have greatly extended human visual capabilities, enabling the observation and understanding of diverse phenomena. Imaging technologies span a broad spectrum of wavelengths from X-ray to radio frequencies and impact research activities and our daily lives. Traditional glass lenses are fabricated through a series of complex processes, while polymers offer versatility and ease of production. However, modern applications often require complex lens assemblies, driving the need for miniaturization and advanced designs with micro- and nanoscale features to surpass the capabilities of traditional fabrication methods. Three-dimensional (3D) printing, or additive manufacturing, presents a solution to these challenges with benefits of rapid prototyping, customized geometries, and efficient production, particularly suited for miniaturized optical imaging devices. Various 3D printing methods have demonstrated advantages over traditional counterparts, yet challenges remain in achieving nanoscale resolutions. Two-photon polymerization lithography (TPL), a nanoscale 3D printing technique, enables the fabrication of intricate structures beyond the optical diffraction limit via the nonlinear process of two-photon absorption within liquid resin. It offers unprecedented abilities, e.g., alignment-free fabrication, micro- and nanoscale capabilities, and rapid prototyping of almost arbitrary complex 3D nanostructures. In this review, we emphasize the importance of the criteria for optical performance evaluation of imaging devices, discuss material properties relevant to TPL, fabrication techniques, and highlight the application of TPL in optical imaging. As the first panoramic review on this topic, it will equip researchers with foundational knowledge and recent advancements of TPL for imaging optics, promoting a deeper understanding of the field. By leveraging on its high-resolution capability, extensive material range, and true 3D processing, alongside advances in materials, fabrication, and design, we envisage disruptive solutions to current challenges and a promising incorporation of TPL in future optical imaging applications.

show abstract

Section: Computing Imaging With Diffractive Optical Neural Networkmentioning

confidence: 99%

Two-photon polymerization lithography for imaging optics

Wang,

Pan,

et al. 2024

Int. J. Extrem. Manuf.

View full text Add to dashboard Cite

show abstract

“…Recently, a wave-based optical neural network, the diffractive neural network (DNN) 14 , was reported. DNNs have demonstrated superior performance in various AI tasks, such as image recognition 14 – 20 , optical computing 21 , phase retrieval 22 , adaptive focusing 23 , and terahertz pulse shaping 24 . In contrast to waveguide-based optical neural networks 25 – 28 , DNNs mimic the human nervous system in three-dimensional (3D) domains.…”

Section: Introductionmentioning

confidence: 99%

Compact eternal diffractive neural network chip for extreme environments

Dong,

Lin,

Chen

et al. 2024

Commun Eng

Self Cite

View full text Add to dashboard Cite

Artificial intelligence applications in extreme environments place high demands on hardware robustness, power consumption, and speed. Recently, diffractive neural networks have demonstrated superb advantages in high-throughput light-speed reasoning. However, the robustness and lifetime of existing diffractive neural networks cannot be guaranteed, severely limiting their compactness and long-term inference accuracy. Here, we have developed a millimeter-scale and robust bilayer-integrated diffractive neural network chip with virtually unlimited lifetime for optical inference. The two diffractive layers with binary phase modulation were engraved on both sides of a quartz wafer. Optical inference of handwritten digital recognition was demonstrated. The results showed that the chip achieved 82% recognition accuracy for ten types of digits. Moreover, the chip demonstrated high-performance stability at high temperatures. The room-temperature lifetime was estimated to be 1.84×1023 trillion years. Our chip satisfies the requirements for diffractive neural network hardware with high robustness, making it suitable for use in extreme environments.

show abstract

DSTANet: learning a dual-stream model for anomaly driving action detection using spatio-temporal and appearance features

Peng,

Hu,

et al. 2024

Multimedia Systems

View full text Add to dashboard Cite

Event-based diffractive neural network chip for dynamic action recognition

Cited by 7 publications

References 53 publications

Two-photon polymerization lithography for imaging optics

Two-photon polymerization lithography for imaging optics

Compact eternal diffractive neural network chip for extreme environments

DSTANet: learning a dual-stream model for anomaly driving action detection using spatio-temporal and appearance features

Contact Info

Product

Resources

About