KLIEP-based Density Ratio Estimation for Semantically Consistent Synthetic to Real Images Adaptation in Urban Traffic Scenes

Abstract

Synthetic data has been applied in many deep learning based computer vision tasks. Limited performance of algorithms trained solely on synthetic data has been approached with domain adaptation techniques such as the ones based on generative adversarial framework. We demonstrate how adversarial training alone can introduce semantic inconsistencies in translated images. To tackle this issue we propose density prematching strategy using KLIEP-based density ratio estimation procedure. Finally, we show that aforementioned strategy improves quality of translated images of underlying method and their usability for the semantic segmentation task in the context of autonomous driving.

Probability density function based data augmentation for deep neural network automatic modulation classification with limited training data

“…A naive method estimates p lim (r) and p suf (r) separately and then obtains Λ(r) by computing the ratio of the estimated PDF. However, density estimation is challenging, particularly for high-dimensional input samples [28]. To improve the estimation accuracy, we introduce the KLIEP approach to estimate Λ(r) directly.…”

“…A naive method estimates $${p_{\lim }}({\mathbf {r}} )$$ and $${p_{{\text{suf}}}}({\mathbf {r}} )$$ separately and then obtains $$\Lambda ({\mathbf {r}} )$$ by computing the ratio of the estimated PDF. However, density estimation is challenging, particularly for high‐dimensional input samples [28]. To improve the estimation accuracy, we introduce the KLIEP approach to estimate $$\Lambda ({\mathbf {r}} )$$ directly.…”