
-
生物通官微
陪你抓住生命科技
跳动的脉搏
使用跨模态层次感知网络评估人工智能生成图像的质量
《IEEE Transactions on Broadcasting》:Assessing AI-Generated Image Quality Using a Cross-Modal Hierarchical Perception Network
【字体: 大 中 小 】 时间:2025年11月22日 来源:IEEE Transactions on Broadcasting 4.8
编辑推荐:
AI-Generated Images (AGIs) quality assessment is crucial for multimedia applications, but existing methods fail to align fine-grained cross-modal semantics and capture diverse quality factors across multiple perceptual levels. This paper proposes CHPNet with MCINet and AHSNet to simulate human hierarchical visual perception. Experiments on three AGIQA databases validate its effectiveness and the code is开源 at specified link.
生物通微信公众号
知名企业招聘