Research map / Architectures and efficiency / CNN design
Efficient Neural Vision Systems Based on Convolutional Image Acquisition
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add a large optical convolution before the perceptron (ablation) | Syntiant perceptron · EMNIST-Digits | accuracy: 95.16% -> 98.29% | Helped |
| Replace linear sensor activation with logarithmic activation (ablation) | OptConv+Perc · EMNIST-Digits | accuracy: 98.29% -> 99.43% | Helped |
| Add a large optical convolution before the perceptron (ablation) | Syntiant perceptron · EMNIST-Letters | accuracy: 76.54% -> 91.92% | Helped |
| Replace linear sensor activation with logarithmic activation (ablation) | OptConv+Perc · EMNIST-Letters | accuracy: 91.92% -> 93.65% | Helped |
| Add a large optical convolution before the perceptron (ablation) | Syntiant perceptron · EMNIST-Balanced | accuracy: 70.21% -> 84.68% | Helped |
| Replace linear sensor activation with logarithmic activation (ablation) | OptConv+Perc · EMNIST-Balanced | accuracy: 84.68% -> 87.69% | Helped |
| Add a large optical convolution before the perceptron (ablation) | Perceptron · InAirGestures test set | accuracy: 97.16% -> 99.94% | Helped |
| Add light-throughput regularization to favor masks with higher L1 norm (ablation) | Optical convolution mask · OCR experiments | accuracy | No clear effect |
| Use optical convolution and logarithmic sensor activation instead of a digital first convolution and linear sensor activation | Syntiant perceptron · EMNIST-Digits | MAdds: 250 | Helped |
| Use optical convolution and logarithmic sensor activation instead of a digital first convolution and linear sensor activation | Syntiant perceptron · EMNIST-Letters | MAdds: 460 | Helped |
Closest papers by meaning
- Deep Optics for Single-Shot High-Dynamic-Range Imaging (CVPR 2020 · similarity 0.87)
- ASP Vision: Optically Computing the First Layer of Convolutional Neural Networks Using Angle Sensitive Pixels (CVPR 2016 · similarity 0.86)
- Neural Tangent Knowledge Distillation for Optical Convolutional Networks (NeurIPS 2025 · similarity 0.84)
- A Physics-Informed Blur Learning Framework for Imaging Systems (CVPR 2025 · similarity 0.84)
- Latent Space Imaging (CVPR 2025 · similarity 0.83)
- Recurrent Partial Kernel Network for Efficient Optical Flow Estimation (AAAI 2024 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.