Oracle Bone Inscriptions (OBI), the earliest systematic writing in China, are crucial for understanding early Chinese civilization. However, many inscriptions remain undeciphered due to limited data, complex glyphs, and weak semantic consistency. Although Generative Adversarial Networks (GAN) and Diffusion Models have introduced new possibilities to the field, most existing methods primarily focus on visual features and lack semantic integration. To address this challenge, we propose DCSD-OBI, a Dual-Conditional Stable Diffusion model that integrates OBI images and modern Chinese text during reverse diffusion to jointly learn structural and semantic features, thereby improving character generation accuracy. To reduce training cost while preserving generative quality, we adopt Low-Rank Adaptation (LoRA) to fine-tune only the U-Net’s cross-attention modules. Furthermore, we replace CLIP with Chinese-CLIP, a version tailored for the Chinese language, to improve cross-modal semantic consistency. Experiments results show DCSD-OBI improves decipherment accuracy by 11%, highlighting its effectiveness and potential for advancing OBI research.
Image inpainting approaches have made considerable progress with the assistance of generative adversarial networks (GANs) recently. However, current inpainting methods are incompetent in handling the cases with large masks and they generally suffer from unreasonable structure. We find that the main reason is the lack of an effective receptive field in the inpainting network. To alleviate this issue, we propose a new two-stage inpainting model called MD-GAN, which is a multi-scale diverse GAN. We inject dense combinations of dilated convolutions in multiple scales of inpainting networks to obtain more effective receptive fields. In fact, the result of inpainting large masks is generally not uniquely deterministic. To this end, we newly propose the multi-scale probabilistic diverse module, which achieves diverse content generation by spatial-adaptive normalization. Meanwhile, the convolutional block attention module is introduced to improve the ability to extract complex features. Perceptual diversity loss is added to enhance diversity. Extensive experiments on benchmark datasets including CelebA-HQ, Places2 and Paris Street View demonstrate that our approach is able to effectively inpaint diverse and structurally reasonable images.