Current self-supervised monocular depth estimation (MDE) approaches encounter performance limitations due to insufficient semantic-spatial knowledge extraction. To address this challenge, we propose Hybrid-depth, a novel framework that systematically integrates foundation models (e.g., CLIP and DINO) to extract visual priors and acquire sufficient contextual information for MDE. Our approach introduces a coarse-to-fine progressive learning framework: 1) Firstly, we aggregate multi-grained features from CLIP (global semantics) and DINO (local spatial details) under contrastive language guidance. A proxy task comparing close-distant image patches is designed to enforce depth-aware feature alignment using text prompts; 2) Next, building on the coarse features, we integrate camera pose information and pixel-wise language alignment to refine depth predictions. This module seamlessly integrates with existing self-supervised MDE pipelines (e.g., Monodepth2, ManyDepth) as a plug-and-play depth encoder, enhancing continuous depth estimation. By aggregating CLIP's semantic context and DINO's spatial details through language guidance, our method effectively addresses feature granularity mismatches. Extensive experiments on the KITTI benchmark demonstrate that our method significantly outperforms SOTA methods across all metrics, which also indeed benefits downstream tasks like BEV perception. Code is available at https://github.com/Zhangwenyao1/Hybrid-depth.

Recent advances in vision-language-action (VLA) models have shown promise in integrating image generation with action prediction to improve generalization and reasoning in robot manipulation. However, existing methods are limited to challenging image-based forecasting, which suffers from redundant information and lacks comprehensive and critical world knowledge, including dynamic, spatial and semantic information. To address these limitations, we propose DreamVLA, a novel VLA framework that integrates comprehensive world knowledge forecasting to enable inverse dynamics modeling, thereby establishing a perception-prediction-action loop for manipulation tasks. Specifically, DreamVLA introduces a dynamic-region-guided world knowledge prediction, integrated with the spatial and semantic cues, which provide compact yet comprehensive representations for action planning. This design aligns with how humans interact with the world by first forming abstract multimodal reasoning chains before acting. To mitigate interference among the dynamic, spatial and semantic information during training, we adopt a block-wise structured attention mechanism that masks their mutual attention, preventing information leakage and keeping each representation clean and disentangled. Moreover, to model the conditional distribution over future actions, we employ a diffusion-based transformer that disentangles action representations from shared latent features. Extensive experiments on both real-world and simulation environments demonstrate that DreamVLA achieves 76.7 success rate on real robot tasks and 4.44 average length on the CALVIN ABC-D benchmarks.

Unsupervised domain adaptation (UDA) in medical image segmentation aims to improve the generalization of deep models by alleviating domain gaps caused by inconsistency across equipment, imaging protocols, and patient conditions. However, existing UDA works remain insufficiently explored and present great limitations: 1) Exhibit cumbersome designs that prioritize aligning statistical metrics and distributions, which limits the model's flexibility and generalization while also overlooking the potential knowledge embedded in unlabeled data; 2) More applicable in a certain domain, lack the generalization capability to handle diverse shifts encountered in clinical scenarios. To overcome these limitations, we introduce MedCon, a unified framework that leverages general unsupervised contrastive pre-training to establish domain connections, effectively handling diverse domain shifts without tailored adjustments. Specifically, it initially explores a general contrastive pre-training to establish domain connections by leveraging the rich prior knowledge from unlabeled images. Thereafter, the pre-trained backbone is fine-tuned using source-based images to ultimately identify per-pixel semantic categories. To capture both intra- and inter-domain connections of anatomical structures, we construct positive-negative pairs from a hybrid aspect of both local and global scales. In this regard, a shared-weight encoder-decoder is employed to generate pixel-level representations, which are then mapped into hyper-spherical space using a non-learnable projection head to facilitate positive pair matching. Comprehensive experiments on diverse medical image datasets confirm that MedCon outperforms previous methods by effectively managing a wide range of domain shifts and showcasing superior generalization capabilities.

Existing domain adaptive medical image segmentation works typically rely on style transfer techniques to mitigate the unexpected domain gap, which inevitably suffers from synthesized artifacts or unreasonable stylization. In this paper, we propose to inject biomedical-related prior knowledge (i.e., intensity and anatomical consistency) as regularization in a prompting manner, bridging the domain gap across modalities. Technically, we develop an efficient scheme called Biomedical-Prompting Dual Augmentation (BPDA) to learn domain-invariant representations by enforcing consistent model predictions across different augmented views. BPDA augments unpaired source and target images from intensity and anatomical aspects in a dual manner, while prompting the framework to fully understand the anatomical structure-invariant features. In this way, our method captures discriminative inherent representations on cross-modality scenarios. Furthermore, we also introduce a Cross-Domain Prototype Denoising (CDPD) in BPDA to refine pseudo-labeling results with the class centroids for a reliable augmentation. Extensive experiments on the cross-modality abdominal and cardiac segmentation benchmarks demonstrate the superiority of our method over state-of-the-art alternatives.

There has been exciting progress in generating images from natural language or layout conditions. However, these methods struggle to faithfully reproduce complex scenes due to the insufficient modeling of multiple objects and their relationships. To address this issue, we leverage the scene graph, a powerful structured representation, for complex image generation. Different from the previous works that directly use scene graphs for generation, we employ the generative capabilities of variational autoencoders and diffusion models in a generalizable manner, compositing diverse disentangled visual clues from scene graphs. Specifically, we first propose a Semantics-Layout Variational AutoEncoder (SL-VAE) to jointly derive (layouts, semantics) from the input scene graph, which allows a more diverse and reasonable generation in a one-to-many mapping. We then develop a Compositional Masked Attention (CMA) integrated with a diffusion model, incorporating (layouts, semantics) with fine-grained attributes as generation guidance. To further achieve graph manipulation while keeping the visual content consistent, we introduce a Multi-Layered Sampler (MLS) for an "isolated" image editing effect. Extensive experiments demonstrate that our method outperforms recent competitors based on text, layout, or scene graph, in terms of generation rationality and controllability.

Disentangled representation learning (DRL) aims to identify and decompose underlying factors behind observations, thus facilitating data perception and generation. However, current DRL approaches often rely on the unrealistic assumption that semantic factors are statistically independent. In reality, these factors may exhibit correlations, which off-the-shelf solutions have yet to properly address. To tackle this challenge, we introduce a bidirectional weighted graph-based framework, to learn factorized attributes and their interrelations within complex data. Specifically, we propose a $\beta$-VAE based module to extract factors as the initial nodes of the graph, and leverage the multimodal large language model (MLLM) to discover and rank latent correlations, thereby updating the weighted edges. By integrating these complementary modules, our model successfully achieves fine-grained, practical and unsupervised disentanglement. Experiments demonstrate our method's superior performance in disentanglement and reconstruction. Furthermore, the model inherits enhanced interpretability and generalizability from MLLMs.

Unsupervised domain adaptation (UDA) aims to learn domain-invariant representations between the labeled source domain and the unlabeled target domain. Existing self- training-based UDA methods use ground truth and pseudo- labels to supervise source data and target data respectively. However, strong supervision in the source domain and pseudo- label noise in the target domain lead to some problems, such as biased predictions and over-fitting. To tackle these issues, we propose a novel Bilateral Knowledge Distillation (BKD) framework for UDA in semantic segmentation, which adopts different knowledge distillation strategies depending on the domain. Specifically, we first introduce a Source-Flow Distillation (SD) to smooth the labels of source images, which weakens the supervision in the source domain. Meanwhile, a Target-Flow Distillation (TD) is designed to extract the inter- class knowledge in the probability map output from the teacher model, which alleviates the influence of pseudo-label noise in the target domain. Considering the class imbalance in semantic segmentation, we further propose an Image-Wise Hard Pixel Mining (HPM) to address this issue without estimating class frequency in the unlabeled target domain. The effectiveness of our framework against existing state-of-the-art methods is demonstrated by extensive experiments on two benchmarks: GTA5-to-Cityscapes and SYNTHIA-to-Cityscapes.