K-Origins: Better Colour Quantification for Neural Networks

TL;DR

K-Origins introduces a novel neural network layer that enhances colour and intensity learning, significantly improving semantic segmentation accuracy in challenging scenarios like low SNR and multi-object colour differentiation.

Contribution

The paper proposes K-Origins, a new neural network layer that improves colour quantification, and demonstrates its effectiveness across various encoder-decoder networks and scenarios.

Findings

K-Origins improves semantic segmentation accuracy in low SNR conditions.

Networks with larger receptive fields benefit from K-Origins for better performance.

Optimal receptive field lengths should exceed object sizes for improved results.

Abstract

K-Origins is a neural network layer designed to improve image-based network performances when learning colour, or intensities, is beneficial. Over 250 encoder-decoder convolutional networks are trained and tested on 16-bit synthetic data, demonstrating that K-Origins improves semantic segmentation accuracy in two scenarios: object detection with low signal-to-noise ratios, and segmenting multiple objects that are identical in shape but vary in colour. K-Origins generates output features from the input features, $\textbf{X}$, by the equation $\textbf{Y}_k = \textbf{X}-\textbf{J}\cdot w_k$ for each trainable parameter $w_k$, where $\textbf{J}$ is a matrix of ones. Additionally, networks with varying receptive fields were trained to determine optimal network depths based on the dimensions of target classes, suggesting that receptive field lengths should exceed object sizes. By ensuring a sufficient receptive field length and incorporating K-Origins, we can achieve better semantic network performance.