Virtual scene tunnel firework generation method and system and storage medium

The invention discloses a virtual scene tunnel firework generation method and system and a storage medium, and belongs to the technical field of deep learning. Tunnel smoke and fire are generated by using a variational auto-encoder, an attention mechanism and a diffusion model according to problems...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: LIU LONGCHEN, XUE ZHIHANG, SHAO QIANQIU, ZHU JUN, BAI HUAN, FU ZHENGZHENG, ZHANG ZONGXI, LUO DONGHUI, HE JUN, LIU FENGLIAN
Format: Patent
Sprache:chi ; eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The invention discloses a virtual scene tunnel firework generation method and system and a storage medium, and belongs to the technical field of deep learning. Tunnel smoke and fire are generated by using a variational auto-encoder, an attention mechanism and a diffusion model according to problems existing in an actual application scene, firstly, an encoder of the variational auto-encoder is used for compressing an input image to a potential space to obtain an input potential variable, and then Gaussian diffusion is performed on the potential variable to obtain a noise-added result; then denoising is carried out through a deep learning network model embedded with an attention module, finally, potential vectors are restored to an image space through a decoder of an auto-encoder, construction of a tunnel smoke and fire image generation network is completed, and a final tunnel smoke and fire generation result is obtained. According to the method provided by the invention, when tunnel smoke and fire generation i