Search papers, labs, and topics across Lattice.
HyPER-GAN, a U-Net-based GAN, is introduced for real-time photorealism enhancement of synthetic images, addressing the limitations of existing methods that suffer from artifacts and high computational costs. The key innovation lies in a hybrid training strategy that combines paired synthetic/enhanced images with matched patches from real-world data. Experiments show HyPER-GAN achieves superior inference speed, visual realism, and semantic robustness compared to state-of-the-art paired image-to-image translation techniques.
Achieve real-time photorealistic image enhancement without sacrificing visual quality or semantic consistency, thanks to a novel hybrid training strategy for GANs.
Generative models are widely employed to enhance the photorealism of synthetic data for training computer vision algorithms. However, they often introduce visual artifacts that degrade the accuracy of these algorithms and require high computational resources, limiting their applicability in real-time training or evaluation scenarios. In this paper, we propose Hybrid Patch Enhanced Realism Generative Adversarial Network (HyPER-GAN), a lightweight image-to-image translation method based on a U-Net-style generator designed for real-time inference. The model is trained using paired synthetic and photorealism-enhanced images, complemented by a hybrid training strategy that incorporates matched patches from real-world data to improve visual realism and semantic consistency. Experimental results demonstrate that HyPER-GAN outperforms state-of-the-art paired image-to-image translation methods in terms of inference latency, visual realism, and semantic robustness. Moreover, it is illustrated that the proposed hybrid training strategy indeed improves visual quality and semantic consistency compared to training the model solely with paired synthetic and photorealism-enhanced images. Code and pretrained models are publicly available for download at: https://github.com/stefanos50/HyPER-GAN