Qingguo Xiao

Zero-shot image style transfer has recently attracted increasing attention, as it enables style manipulation without additional model optimization or parameter tuning. However, most existing approaches still rely on latent inversion, where introduced reconstruction errors may accumulate and gradually degrade content structures. Moreover, current style transfer methods still struggle to balance style strength and content preservation due to inadequate guidance. To address these issues, this paper proposes a novel zero-shot image style transfer method based on conditional inversion-free flow. Built upon pre-trained Flow Matching models, our method eliminates the inversion by constructing a style-transfer trajectory ODE from a gradient-injected discrepancy between content-conditioned and style-conditioned velocity fields. To further enhance style representation, we introduce a conditional guidance mechanism driven by a pre-trained CLIP model. Specifically, a style loss is introduced to align the Gram matrix statistics of CLIP features with those of the reference style; a semantic consistency constraint based on CLIP embedding cosine distance preserves high-level semantics. The dual guidance terms are incorporated during the inversion-free editing stage through loss gradients, guiding the generation trajectory without model fine-tuning or access to internal model features. Extensive quantitative and qualitative experiments demonstrate that our method outperforms state-of-the-art approaches, providing a better balance between style consistency and content preservation.

Weakly supervised semantic segmentation builds a semantic segmentation model with only image-level annotations, which provide only categorical information without localization details. This limitation leads to persistent challenges related to class co-occurrence. Traditional methods often address this issue by incorporating external data or employing data augmentation techniques. In contrast, we benefit from the large and diverse amount of data seen during SAM pre-training to exploit its potential recognition ability and solve the class co-occurrence problem with only the pre-trained model. Furthermore, existing SAM-based methods typically adopt SAM to produce pseudo-labels, which are then employed to train separate segmentation networks without employing SAM as a backbone for direct segmentation with image-level labels. In addition, we found that although the SAM-based approach can solve the class co-occurrence problem, it still suffers from full target leakage. Based on these observations, we propose a one-stage, dual-teachers-one-student network architecture, termed Co2SAM. Specifically, we utilize Focal loss, Contrast loss, Dice loss, and Template loss, and fine-tune the image encoder with LoRA. Experimental results substantiate the effectiveness of our Co2SAM. The code is available at https://github.com/chunmengliu666/Co2SAM.

Image inpainting is a technique that aims to fill in the missing regions with visually plausible content. However, an opposite idea, which is painting outside images, receives little work. In this study, we investigate the issue of image outpainting. Considering that the model needs better prediction ability as there is less neighboring information in image outpainting, the study proposes a novel image outpainting architecture that can obtain both deep model performance and detailed information. To fully take advantage of residual learning, dense residual (DR) learning is proposed and the image generative network is built on DR. To avoid losing subtle information caused by downsampling in encoder-decoder, shortcuts are added for transferring previous knowledge. Different from vanilla U-Net, we propose a skip method of the semi-complete form. Experimental results show that the proposed method achieves excellent performance.