Existing text-to-image models still struggle to generate images of multiple objects, especially in handling their spatial positions, relative sizes, overlapping, and attribute bindings. In this paper, we develop a training-free Multimodal-LLM agent (MuLan) to address these challenges by progressive multi-object generation with planning and feedback control, like a human painter. MuLan harnesses a large language model (LLM) to decompose a prompt to a sequence of sub-tasks, each generating only one object conditioned on previously generated objects by stable diffusion. Unlike existing LLM-grounded methods, MuLan only produces a high-level plan at the beginning while the exact size and location of each object are determined by an LLM and attention guidance upon each sub-task. Moreover, MuLan adopts a vision-language model (VLM) to provide feedback to the image generated in each sub-task and control the diffusion model to re-generate the image if it violates the original prompt. Hence, each model in every step of MuLan only needs to address an easy sub-task it is specialized for. We collect 200 prompts containing multi-objects with spatial relationships and attribute bindings from different benchmarks to evaluate MuLan. The results demonstrate the superiority of MuLan in generating multiple objects over baselines. The code is available on https://github.com/measure-infinity/mulan-code.
We present Multi-HMR, a strong single-shot model for multi-person 3D human mesh recovery from a single RGB image. Predictions encompass the whole body, i.e, including hands and facial expressions, using the SMPL-X parametric model and spatial location in the camera coordinate system. Our model detects people by predicting coarse 2D heatmaps of person centers, using features produced by a standard Vision Transformer (ViT) backbone. It then predicts their whole-body pose, shape and spatial location using a new cross-attention module called the Human Prediction Head (HPH), with one query per detected center token, attending to the entire set of features. As direct prediction of SMPL-X parameters yields suboptimal results, we introduce CUFFS; the Close-Up Frames of Full-Body Subjects dataset, containing humans close to the camera with diverse hand poses. We show that incorporating this dataset into training further enhances predictions, particularly for hands, enabling us to achieve state-of-the-art performance. Multi-HMR also optionally accounts for camera intrinsics, if available, by encoding camera ray directions for each image token. This simple design achieves strong performance on whole-body and body-only benchmarks simultaneously. We train models with various backbone sizes and input resolutions. In particular, using a ViT-S backbone and $448\times448$ input images already yields a fast and competitive model with respect to state-of-the-art methods, while considering larger models and higher resolutions further improve performance.
Deep learning-based methods have demonstrated encouraging results in tackling the task of panoramic image inpainting. However, it is challenging for existing methods to distinguish valid pixels from invalid pixels and find suitable references for corrupted areas, thus leading to artifacts in the inpainted results. In response to these challenges, we propose a panoramic image inpainting framework that consists of a Face Generator, a Cube Generator, a side branch, and two discriminators. We use the Cubemap Projection (CMP) format as network input. The generator employs gated convolutions to distinguish valid pixels from invalid ones, while a side branch is designed utilizing contextual reconstruction (CR) loss to guide the generators to find the most suitable reference patch for inpainting the missing region. The proposed method is compared with state-of-the-art (SOTA) methods on SUN360 Street View dataset in terms of PSNR and SSIM. Experimental results and ablation study demonstrate that the proposed method outperforms SOTA both quantitatively and qualitatively.
Current image manipulation primarily centers on static manipulation, such as replacing specific regions within an image or altering its overall style. In this paper, we introduce an innovative dynamic manipulation task, subject repositioning. This task involves relocating a user-specified subject to a desired position while preserving the image's fidelity. Our research reveals that the fundamental sub-tasks of subject repositioning, which include filling the void left by the repositioned subject, reconstructing obscured portions of the subject and blending the subject to be consistent with surrounding areas, can be effectively reformulated as a unified, prompt-guided inpainting task. Consequently, we can employ a single diffusion generative model to address these sub-tasks using various task prompts learned through our proposed task inversion technique. Additionally, we integrate pre-processing and post-processing techniques to further enhance the quality of subject repositioning. These elements together form our SEgment-gEnerate-and-bLEnd (SEELE) framework. To assess SEELE's effectiveness in subject repositioning, we assemble a real-world subject repositioning dataset called ReS. Our results on ReS demonstrate the quality of repositioned image generation.
This paper introduces a public dataset of 1.4 million procedurally-generated bicycle designs represented parametrically, as JSON files, and as rasterized images. The dataset is created through the use of a rendering engine which harnesses the BikeCAD software to generate vector graphics from parametric designs. This rendering engine is discussed in the paper and also released publicly alongside the dataset. Though this dataset has numerous applications, a principal motivation is the need to train cross-modal predictive models between parametric and image-based design representations. For example, we demonstrate that a predictive model can be trained to accurately estimate Contrastive Language-Image Pretraining (CLIP) embeddings from a parametric representation directly. This allows similarity relations to be established between parametric bicycle designs and text strings or reference images. Trained predictive models are also made public. The dataset joins the BIKED dataset family which includes thousands of mixed-representation human-designed bicycle models and several datasets quantifying design performance. The code and dataset can be found at: https://github.com/Lyleregenwetter/BIKED_multimodal/tree/main
In the field of Medical Imaging, extensive research has been dedicated to leveraging its potential in uncovering critical diagnostic features in patients. Artificial Intelligence (AI)-driven medical diagnosis relies on sophisticated machine learning and deep learning models to analyze, detect, and identify diseases from medical images. Despite the remarkable performance of these models, characterized by high accuracy, they grapple with trustworthiness issues. The introduction of a subtle perturbation to the original image empowers adversaries to manipulate the prediction output, redirecting it to other targeted or untargeted classes. Furthermore, the scarcity of publicly available medical images, constituting a bottleneck for reliable training, has led contemporary algorithms to depend on pretrained models grounded on a large set of natural images -- a practice referred to as transfer learning. However, a significant {\em domain discrepancy} exists between natural and medical images, which causes AI models resulting from transfer learning to exhibit heightened {\em vulnerability} to adversarial attacks. This paper proposes a {\em domain assimilation} approach that introduces texture and color adaptation into transfer learning, followed by a texture preservation component to suppress undesired distortion. We systematically analyze the performance of transfer learning in the face of various adversarial attacks under different data modalities, with the overarching goal of fortifying the model's robustness and security in medical imaging tasks. The results demonstrate high effectiveness in reducing attack efficacy, contributing toward more trustworthy transfer learning in biomedical applications.
The availability of city-scale Lidar maps enables the potential of city-scale place recognition using mobile cameras. However, the city-scale Lidar maps generally need to be compressed for storage efficiency, which increases the difficulty of direct visual place recognition in compressed Lidar maps. This paper proposes VOLoc, an accurate and efficient visual place recognition method that exploits geometric similarity to directly query the compressed Lidar map via the real-time captured image sequence. In the offline phase, VOLoc compresses the Lidar maps using a \emph{Geometry-Preserving Compressor} (GPC), in which the compression is reversible, a crucial requirement for the downstream 6DoF pose estimation. In the online phase, VOLoc proposes an online Geometric Recovery Module (GRM), which is composed of online Visual Odometry (VO) and a point cloud optimization module, such that the local scene structure around the camera is online recovered to build the \emph{Querying Point Cloud} (QPC). Then the QPC is compressed by the same GPC, and is aggregated into a global descriptor by an attention-based aggregation module, to query the compressed Lidar map in the vector space. A transfer learning mechanism is also proposed to improve the accuracy and the generality of the aggregation network. Extensive evaluations show that VOLoc provides localization accuracy even better than the Lidar-to-Lidar place recognition, setting up a new record for utilizing the compressed Lidar map by low-end mobile cameras. The code are publicly available at https://github.com/Master-cai/VOLoc.
Computer vision-based analysis of street view imagery has transformative impacts on environmental assessments. Interactive web services, particularly Google Street View, play an ever-important role in making imagery data ubiquitous. Despite the technical ease of harnessing millions of Google Street View images, this article questions the current practices in using this proprietary data source. Our concern lies with Google's terms of service, which prohibit bulk image downloads and the generation of street view image-based indices. To reconcile the challenge of advancing society through groundbreaking research while maintaining data license agreements and legal integrity, it is crucial to adhere to open data principles and utilize open image sources for future research.
Diffusion models are widely used in image generation because they can generate high-quality and realistic samples. This is in contrast to generative adversarial networks (GANs) and variational autoencoders (VAEs), which have some limitations in terms of image quality.We introduce the diffusion model to the precipitation forecasting task and propose a short-term precipitation nowcasting with condition diffusion model based on historical observational data, which is referred to as SRNDiff. By incorporating an additional conditional decoder module in the denoising process, SRNDiff achieves end-to-end conditional rainfall prediction. SRNDiff is composed of two networks: a denoising network and a conditional Encoder network. The conditional network is composed of multiple independent UNet networks. These networks extract conditional feature maps at different resolutions, providing accurate conditional information that guides the diffusion model for conditional generation.SRNDiff surpasses GANs in terms of prediction accuracy, although it requires more computational resources.The SRNDiff model exhibits higher stability and efficiency during training than GANs-based approaches, and generates high-quality precipitation distribution samples that better reflect future actual precipitation conditions. This fully validates the advantages and potential of diffusion models in precipitation forecasting, providing new insights for enhancing rainfall prediction.
The fusion of a low-spatial-resolution hyperspectral image (LR-HSI) with a high-spatial-resolution multispectral image (HR-MSI) has garnered increasing research interest. However, most fusion methods solely focus on the fusion algorithm itself and overlook the degradation models, which results in unsatisfactory performance in practical scenarios. To fill this gap, we propose physics-inspired degradation models (PIDM) to model the degradation of LR-HSI and HR-MSI, which comprises a spatial degradation network (SpaDN) and a spectral degradation network (SpeDN). SpaDN and SpeDN are designed based on two insights. First, we employ spatial warping and spectral modulation operations to simulate lens aberrations, thereby introducing non-uniformity into the spatial and spectral degradation processes. Second, we utilize asymmetric downsampling and parallel downsampling operations to separately reduce the spatial and spectral resolutions of the images, thus ensuring the matching of spatial and spectral degradation processes with specific physical characteristics. Once SpaDN and SpeDN are established, we adopt a self-supervised training strategy to optimize the network parameters and provide a plug-and-play solution for fusion methods. Comprehensive experiments demonstrate that our proposed PIDM can boost the fusion performance of existing fusion methods in practical scenarios.