Posted 2025-01-06Updated 2025-02-19Notea few seconds read (About 34 words) visitsCLIP https://blog.csdn.net/h661975/article/details/135116957 CLIP 本质上是全局图像嵌入，不利于像素对齐特征提取。 CLIPhttp://chen-yulin.github.io/2025/01/06/[OBS]Reconstruct Anything-Segmentation-CLIP多模态预训练模型/AuthorChen YulinPosted on2025-01-06Updated on2025-02-19Licensed under#CVMulti-modal
