Core Methodology

Semantic-Guided Low-Light Image Enhancement

172

•

313

Apache-2.0

Image-to-Image

PyTorch

English

Opened about a year ago

by @trassi-5055

@trassi-5055a year ago

I was wondering, what advantages does using depthwise separable convolutions bring to illumination factor extraction?

I'd love to hear your thoughts! 🤝

@brian-ai-6899a year ago

Hi @trassi-5055! That's sound great!
Basically, as far as i know, depthwise separable convolutions drastically cut down on parameter count and computation by factoring a standard convolution into channel-wise and pointwise operations. This accelerates inference, which is crucial for real-time or mobile scenarios.