The incorporation of high‐performance optoelectronic devices into photonic neuromorphic processors can substantially accelerate computationally intensive matrix multiplication operations in machine learning (ML) algorithms. However, the conventional designs of individual devices and system are largely disconnected, and the system optimization is limited to the manual exploration of a small design space. Here, a device‐system end‐to‐end design methodology is reported to optimize a free‐space optical general matrix multiplication (GEMM) hardware accelerator by engineering a spatially reconfigurable array made from chalcogenide phase change materials. With a highly parallelized integrated hardware emulator with experimental information, the design of unit device to directly optimize GEMM calculation accuracy is achieved by exploring a large parameter space through reinforcement learning algorithms, including deep Q‐learning neural network, Bayesian optimization, and their cascaded approach. The algorithm‐generated physical quantities show a clear correlation between system performance metrics and device specifications. Furthermore, physics‐aware training approaches are employed to deploy optimized hardware to the tasks of image classification, materials discovery, and a closed‐loop design of optical ML accelerators. The demonstrated framework offers insights into the end‐to‐end and co‐design of optoelectronic devices and systems with reduced human supervision and domain knowledge barriers.
Differentiable Compound Optics and Processing Pipeline Optimization for End-to-end Camera Design
Most modern commodity imaging systems we use directly for photography—or indirectly rely on for downstream applications—employ optical systems of multiple lenses that must balance deviations from perfect optics, manufacturing constraints, tolerances, cost, and footprint. Although optical designs often have complex interactions with downstream image processing or analysis tasks, today’s compound optics are designed in isolation from these interactions. Existing optical design tools aim to minimize optical aberrations, such as deviations from Gauss’ linear model of optics, instead of application-specific losses, precluding joint optimization with hardware image signal processing (ISP) and highly parameterized neural network processing. In this article, we propose an optimization method for compound optics that lifts these limitations. We optimize entire lens systems jointly with hardware and software image processing pipelines, downstream neural network processing, and application-specific end-to-end losses. To this end, we propose a learned, differentiable forward model for compound optics and an alternating proximal optimization method that handles function compositions with highly varying parameter dimensions for optics, hardware ISP, and neural nets. Our method integrates seamlessly atop existing optical design tools, such as Zemax . We can thus assess our method across many camera system designs and end-to-end applications. We validate our approach in an automotive camera optics setting—together with hardware ISP post processing and detection—outperforming classical optics designs for automotive object detection and traffic light state detection. For human viewing tasks, we optimize optics and processing pipelines for dynamic outdoor scenarios and dynamic low-light imaging. We outperform existing compartmentalized design or fine-tuning methods qualitatively and quantitatively, across all domain-specific applications tested.
more »
« less
- Award ID(s):
- 2047359
- PAR ID:
- 10301671
- Date Published:
- Journal Name:
- ACM Transactions on Graphics
- Volume:
- 40
- Issue:
- 2
- ISSN:
- 0730-0301
- Page Range / eLocation ID:
- 1 to 19
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
more » « less
Abstract -
The rapid progress in intelligent vehicle technology has led to a significant reliance on computer vision and deep neural networks (DNNs) to improve road safety and driving experience. However, the image signal processing (ISP) steps required for these networks, including demosaicing, color correction, and noise reduction, increase the overall processing time and computational resources. To address this, our paper proposes an improved version of the Faster R-CNN algorithm that integrates camera parameters into raw image input, reducing dependence on complex ISP steps while enhancing object detection accuracy. Specifically, we introduce additional camera parameters, such as ISO speed rating, exposure time, focal length, and F-number, through a custom layer into the neural network. Further, we modify the traditional Faster R-CNN model by adding a new fully connected layer, combining these parameters with the original feature maps from the backbone network. Our proposed new model, which incorporates camera parameters, has a 4.2% improvement in mAP@[0.5,0.95] compared to the traditional Faster RCNN model for object detection tasks on raw image data.more » « less
-
FlowCam: Training generalizable 3D radiance fields without camera poses via pixel-aligned scene flowReconstruction of 3D neural fields from posed images has emerged as a promising method for self-supervised representation learning. The key challenge preventing the deployment of these 3D scene learners on large-scale video data is their dependence on precise camera poses from structure-from-motion, which is prohibitively expensive to run at scale. We propose a method that jointly reconstructs camera poses and 3D neural scene representations online and in a single forward pass. We estimate poses by first lifting frame-to-frame optical flow to 3D scene flow via differentiable rendering, preserving locality and shift-equivariance of the image processing backbone. SE(3) camera pose estimation is then performed via a weighted least-squares fit to the scene flow field. This formulation enables us to jointly supervise pose estimation and a generalizable neural scene representation via re-rendering the input video, and thus, train end-to-end and fully self-supervised on real-world video datasets. We demonstrate that our method performs robustly on diverse, real-world video, notably on sequences traditionally challenging to optimization-based pose estimation techniques.more » « less
-
FlowCam: Training generalizable 3D radiance fields without camera poses via pixel-aligned scene flowReconstruction of 3D neural fields from posed images has emerged as a promising method for self-supervised representation learning. The key challenge preventing the deployment of these 3D scene learners on large-scale video data is their dependence on precise camera poses from structure-from-motion, which is prohibitively expensive to run at scale. We propose a method that jointly reconstructs camera poses and 3D neural scene representations online and in a single forward pass. We estimate poses by first lifting frame-to-frame optical flow to 3D scene flow via differentiable rendering, preserving locality and shift-equivariance of the image processing backbone. SE(3) camera pose estimation is then performed via a weighted least-squares fit to the scene flow field. This formulation enables us to jointly supervise pose estimation and a generalizable neural scene representation via re-rendering the input video, and thus, train end-to-end and fully self-supervised on real-world video datasets. We demonstrate that our method performs robustly on diverse, real-world video, notably on sequences traditionally challenging to optimization-based pose estimation techniques.more » « less
-
more » « less
The explosive growth in computation and energy cost of artificial intelligence has spurred interest in alternative computing modalities to conventional electronic processors. Photonic processors, which use photons instead of electrons, promise optical neural networks with ultralow latency and power consumption. However, existing optical neural networks, limited by their designs, have not achieved the recognition accuracy of modern electronic neural networks. In this work, we bridge this gap by embedding parallelized optical computation into flat camera optics that perform neural network computations during capture, before recording on the sensor. We leverage large kernels and propose a spatially varying convolutional network learned through a low-dimensional reparameterization. We instantiate this network inside the camera lens with a nanophotonic array with angle-dependent responses. Combined with a lightweight electronic back-end of about 2K parameters, our reconfigurable nanophotonic neural network achieves 72.76% accuracy on CIFAR-10, surpassing AlexNet (72.64%), and advancing optical neural networks into the deep learning era.
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.1145/3446791
