zz123ff/sd1_4_Fine-tuning_with_white_blood_cells

DreamBooth model for the WBC38 category concept trained by zz123ff on the zz123ff/WBC_38category-images dataset.

This is a Stable Diffusion model fine-tuned on the WBC38 category concept with DreamBooth. It leverages CLIP for robust text encoding and Janus-Pro1B for generating detailed image descriptions, providing enhanced multi-modal conditioning for the diffusion process. It can be used by modifying the `instanceprompt`: "A high-resolution microscopy image of a white blood cell, exhibiting clear morphological features and detailed textures.".

使用WBC38类别概念的模型，由 zz123ff 基于 zz123ff/WBC_38category-images 数据集训练。

这是一个基于WBC38类别图像数据集微调得到的稳定扩散模型，采用DreamBooth进行训练。模型结合了CLIP用于文本编码和Janus-Pro1B用于图像描述(设置最大生成token长度为70)生成，从而为扩散过程提供了更准确的多模态条件信息。您可以通过修改 `instanceprompt` 来使用此模型："A high-resolution microscopy image of a white blood cell, exhibiting clear morphological features and detailed textures."。

Description / 描述

This model was fine-tuned on the WBC38 category images dataset for medical cell classification tasks. During training, CLIP was used for extracting robust text features, and Janus-Pro_1B was employed to generate detailed image descriptions, enhancing the model's conditioning.

该模型基于WBC38类别图像数据集进行了微调，主要用于医学细胞分类任务。在训练过程中，采用CLIP进行文本特征提取，并使用Janus-Pro_1B生成详细的图像描述(设置最大生成token长度为70)，从而为模型提供了更好的多模态条件信息。