High-quality and diverse datasets are crucial for supervised learning in industrial defect detection, yet collecting such data remains challenging. Synthetic data generation offers a promising solution for data augmentation, but traditional methods struggle with controlling key semantic features, leading to domain gaps that impair downstream performance. To address these challenges, we introduce DD-Aug, a novel data augmentation pipeline that integrates Cinema 4-D (C4D) and stable diffusion (SD) to generate photorealistic synthetic images. C4D is used to model key semantic features, such as contours, textures, and lighting conditions, while SD performs style transfer to increase the realism of generated images and reduce the domain gap. DD-Aug translates human visual knowledge into synthetic images, offering precise control over semantic features while adhering to physical rules and reducing the need for complex prompt engineering. This approach enhances generalization, making it particularly suitable for challenging industrial imaging conditions. By balancing semantic consistency and data diversity, DD-Aug significantly improves the quality of synthetic data. Our evaluations on the deformable defect bearing dataset (NBBD) demonstrate that DD-Aug surpasses other generative augmentation methods, yielding a 2.5%–10% improvement in detection accuracy across multiple advanced object detectors compared to models without augmentation. These results demonstrate DD-Aug's potential to significantly advance industrial defect detection through improved synthetic data quality and performance.

3D point cloud data are widely used in many fields as they can be easily obtained and contain rich semantic information. Recently, weakly supervised segmentation has attracted lots of attention because it only requires very few labels, thus reducing time-consuming and expensive data annotation efforts for huge amounts of point cloud data. Existing approaches typically adopt softmax scores from the last layer as the confidence for selecting high-confident point predictions. However, such approaches can ignore the potential value of a large number of low-confidence point predictions under traditional metrics. In this work, we propose an Uncertainty-guided Contrastive Learning (UCL) framework for weakly supervised point cloud segmentation. A novel uncertainty metric based on prototype entropy is presented to estimate the reliability of model predictions. With this metric, we propose a negative contrastive learning module exploiting negative pseudo-labels of predictions with low reliability and an active contrastive learning module enhancing feature learning of segmentation models by predictions with high reliability. We also propose a generic multi-scale feature perturbation method to expand a wider perturbation space. Extensive experimental results on indoor and outdoor point cloud datasets demonstrate that the proposed method achieves competitive performance.

Surface defect segmentation is the basis of industrial production, where pixel-level defect prediction plays a vital role in many applications. As a new trend, few-shot defect detection aims to recognize defects by combining information from a few defect samples. However, many traditional few-shot methods do not consider patch-level correlations between input images and memory banks. This work introduces a novel representation framework consisting hierarchical normal and defect memory banks for few-shot surface defect detection. We propose an adaptive feature bank updating scheme that dynamically integrates frequent features while discarding rare ones. We design a fine-grain feature replacement module rooted in a matching attention mechanism to highlight ambiguous regions for input images, and use fused features to identify overall defects. On benchmark data sets, our strategy can obtain in-depth insights from multiple scales, which showcase superiority in generalization for unidentified anomalies better than others in experiments. The experimental results for the 10-shot scenarios demonstrate that the proposed method achieves an average pixel AUC of 88.6%, 93.7%, and 98.7% on three benchmark data sets, respectively, and with more genuine and pseudo anomalous samples, the proposed approach can achieve 98.5% and 95.5% in image AUC and pixel F1 score respectively, outperforming the state-of-the-art with 1.2% and 24% in each metric.

Three-dimensional (3D) pose estimation has been widely used in many three-dimensional human motion analysis applications, where inertia-based path estimation is gradually being adopted. Systems based on commercial inertial measurement units (IMUs) usually rely on dense and complex wearable sensors and time-consuming calibration, causing intrusions to the subject and hindering free body movement. The sparse IMUs-based method has drawn research attention recently. Existing sparse IMUs-based three-dimensional pose estimation methods use neural networks to obtain human poses from temporal feature information. However, these methods still suffer from issues, such as body shaking, body tilt, and movement ambiguity. This paper presents an approach to improve three-dimensional human pose estimation by fusing temporal and spatial features. Based on a multistage encoder–decoder network, a temporal convolutional encoder and human kinematics regression decoder were designed. The final three-dimensional pose was predicted from the temporal feature information and human kinematic feature information. Extensive experiments were conducted on two benchmark datasets for three-dimensional human pose estimation. Compared to state-of-the-art methods, the mean per joint position error was decreased by 13.6% and 19.4% on the total capture and DIP-IMU datasets, respectively. The quantitative comparison demonstrates that the proposed temporal information and human kinematic topology can improve pose accuracy.

Background Model-based 3D pose estimation has been widely used in many 3D human motion analysis applications, in which vision-based and inertial-based are two distinct lines. Multi-view images in a vision-based markerless capture system provide essential data for motion analysis, but erroneous estimates still occur due to ambiguities, occlusion, or noise in images. Besides, the multi-view setting is hard for the application in the wild. Although inertial measurement units (IMUs) can obtain accurate direction without occlusion, they are usually susceptible to magnetic field interference and drifts. Hybrid motion capture has drawn the attention of researchers in recent years. Existing 3D pose estimation methods jointly optimize the parameters of the 3D pose by minimizing the discrepancy between the image and IMU data. However, these hybrid methods still suffer from the issues such as complex peripheral devices, sensitivity to initialization, and slow convergence. Methods This article presents an approach to improve 3D human pose estimation by fusing a single image with sparse inertial measurement units (IMUs). Based on a dual-stream feature extract network, we design a model-attention network with a residual module to closely couple the dual-modal feature from a static image and sparse inertial measurement units. The final 3D pose and shape parameters are directly obtained by a regression strategy. Results Extensive experiments are conducted on two benchmark datasets for 3D human pose estimation. Compared to state-of-the-art methods, the per vertex error (PVE) of human mesh reduces by 9.4 mm on Total Capture dataset and the mean per joint position error (MPJPE) reduces by 7.8 mm on the Human3.6M dataset. The quantitative comparison demonstrates that the proposed method could effectively fuse sparse IMU data and images and improve pose accuracy.