Creating recipe images is a key challenge in food computing, with applications in culinary education and multimodal recipe assistants. However, existing datasets lack fine-grained alignment between recipe goals, step-wise instructions, and visual content. We present RecipeGen, the first large-scale, real-world benchmark for recipe-based Text-to-Image (T2I), Image-to-Video (I2V), and Text-to-Video (T2V) generation. RecipeGen contains 26,435 recipes, 196,724 images, and 4,491 videos, covering diverse ingredients, cooking procedures, styles, and dish types. We further propose domain-specific evaluation metrics to assess ingredient fidelity and interaction modeling, benchmark representative T2I, I2V, and T2V models, and provide insights for future recipe generation models. Project page is available at https://wenbin08.github.io/RecipeGen.

Cooking is a sequential and visually grounded activity, where each step such as chopping, mixing, or frying carries both procedural logic and visual semantics. While recent diffusion models have shown strong capabilities in text-to-image generation, they struggle to handle structured multi-step scenarios like recipe illustration. Additionally, current recipe illustration methods are unable to adjust to the natural variability in recipe length, generating a fixed number of images regardless of the actual instructions structure. To address these limitations, we present CookAnything, a flexible and consistent diffusion-based framework that generates coherent, semantically distinct image sequences from textual cooking instructions of arbitrary length. The framework introduces three key components: (1) Step-wise Regional Control (SRC), which aligns textual steps with corresponding image regions within a single denoising process; (2) Flexible RoPE, a step-aware positional encoding mechanism that enhances both temporal coherence and spatial diversity; and (3) Cross-Step Consistency Control (CSCC), which maintains fine-grained ingredient consistency across steps. Experimental results on recipe illustration benchmarks show that CookAnything performs better than existing methods in training-based and training-free settings. The proposed framework supports scalable, high-quality visual synthesis of complex multi-step instructions and holds significant potential for broad applications in instructional media, and procedural content creation. More details are at https://github.com/zhangdaxia22/CookAnything.

With the rapid advancement of diffusion models, text-to-image generation has achieved significant progress in image resolution, detail fidelity, and semantic alignment, particularly with models like Stable Diffusion 3.5, Stable Diffusion XL, and FLUX.1. However, generating emotionally expressive and abstract artistic images remains a major challenge, largely due to the lack of large-scale, fine-grained emotional datasets. To address this gap, we present the EmoArt Dataset-one of the most comprehensive emotion-annotated art datasets to date. It contains 132,664 artworks across 56 painting styles (e.g., Impressionism, Expressionism, Abstract Art), offering rich stylistic and cultural diversity. Each image includes structured annotations: objective scene descriptions, five key visual attributes (brushwork, composition, color, line, light), binary arousal-valence labels, twelve emotion categories, and potential art therapy effects. Using EmoArt, we systematically evaluate popular text-to-image diffusion models for their ability to generate emotionally aligned images from text. Our work provides essential data and benchmarks for emotion-driven image synthesis and aims to advance fields such as affective computing, multimodal learning, and computational art, enabling applications in art therapy and creative design. The dataset and more details can be accessed via the following link: https://zhiliangzhang.github.io/EmoArt-130k/

Drug combination therapy is a promising strategy for managing complex and co-existing diseases. However, drug-drug interactions (DDIs) can result in unexpected adverse effects, making it crucial to understand such interactions to prevent adverse drug reactions and develop new therapeutic strategies. Current DDI annotation methods heavily rely on atom-level graph structural features, overlooking valuable drug contextual representations within medical literature. Additionally, these methods are typically designed for a specific task, limiting their scalability to diverse medical scenarios. To address these limitations, we propose TEmbed-DDI, a novel framework that leverages contextual representations and pre-trained large language model embeddings to enhance feature extraction for DDI annotations. Specifically, we retrieve meaningful contextual texts for each drug to enrich semantic features and adopt pre-trained large language model embeddings to capture rich features from these long-range contextual representations. TEmbed-DDI is the first framework to incorporate LLM-powered embeddings for medical interaction annotations. Furthermore, a bidirectional neural network is integrated into TEmbed-DDI for the integrative Western and traditional Chinese medicine DDI annotation tasks. Comparative results demonstrate that TEmbed-DDI achieves state-of-the-art performance, with the highest AUC scores of 0.992 and 0.95 on the Western CHCH and DEEP interaction annotation benchmarks. Even for the newly constructed Traditional Chinese Medicine (TCM) DDI annotation benchmark, TEmbed-DDI consistently exhibits outstanding generalization capability, achieving an AUC of 0.956. Moreover, case studies further validate TEmbed-DDI's capability to annotate previously unknown interactions. These findings suggest that TEmbed-DDI can serve as a valuable tool in annotating previously unknown drug combinations for real-world applications, facilitating the development of efficacious therapies. Furthermore, as the first framework combining traditional Chinese medicine into DDI annotation tasks, its adaptability highlights the potential in supporting cross-domain medical research. TEmbed-DDI's design principles can inspire the development of flexible LLM-powered frameworks for drug combination discovery in the future.

This study focuses on the task of single tree segmentation in urban environments and improves the PDE-Net model to enhance its ability to process complex features. The original PDE-Net model has limitations when dealing with trees in a large area of land use. In this study, we propose an improved model, PDEC-Net, which combines the original model with the Encoder-Decoder structure, and introduces the Depthwise Separable Convolution layer to better capture the global and local relationships of point clouds. The experimental results show that the PDEC-Net model outperforms the original model in evaluation indicators such as Pre, Recall, and F1 score. Compared with the original PDE-Net, the PDEC-Net has increased Pre, Recall, and F1 score by 4.2%, 0.6%, and 3.2%, respectively. Meanwhile, the improved model also shows better adaptability when dealing with large-scale test sets. Therefore, the PDEC-Net model demonstrates stronger performance in single tree segmentation tasks, providing more powerful support for urban planning, management, and environmental monitoring.