Search for: All records

ObjectiveTo identify lifting actions and count the number of lifts performed in videos based on robust class prediction and a streamlined process for reliable real-time monitoring of lifting tasks. BackgroundTraditional methods for recognizing lifting actions often rely on deep learning classifiers applied to human motion data collected from wearable sensors. Despite their high performance, these methods can be difficult to implement on systems with limited hardware resources. MethodThe proposed method follows a five-stage process: (1) BlazePose, a real-time pose estimation model, detects key joints of the human body. (2) These joints are preprocessed by smoothing, centering, and scaling techniques. (3) Kinematic features are extracted from the preprocessed joints. (4) Video frames are classified as lifting or nonlifting using rank-altered kinematic feature pairs. (5) A lifting counting algorithm counts the number of lifts based on the class predictions. ResultsNine rank-altered kinematic feature pairs are identified as key pairs. These pairs were used to construct an ensemble classifier, which achieved 0.89 or above in classification metrics, including accuracy, precision, recall, and F1 score. This classifier showed an accuracy of 0.90 in lifting counting and a latency of 0.06 ms, which is at least 12.5 times faster than baseline classifiers. ConclusionThis study demonstrates that computer vision-based kinematic features could be adopted to effectively and efficiently recognize lifting actions. ApplicationThe proposed method could be deployed on various platforms, including mobile devices and embedded systems, to monitor lifting tasks in real-time for the proactive prevention of work-related low-back injuries. 

Workers performing repetitive lifting tasks are at high risk of developing low-back work-related musculoskeletal disorders. While the Revised NIOSH Lifting Equation (RNLE) is a widely used tool for evaluating lifting-related risks, its reliance on manual measurement limits its scalability and efficiency. This study proposes a computer vision-based framework that automates RNLE computation using video data. The method integrates three key stages: (1) pose estimation to extract 3D joint coordinates, (2) lifting action recognition via kinematic features and a k-TSP classifier, and (3) estimation of RNLE multipliers from joint data. Applied to 40 lifting trials with motion capture-based ground-truth, the system achieved a coefficient of determination of 0.82 and a mean absolute error of 2.72 kg in estimating recommended weight limits. These findings demonstrate the potential of computer vision to automate ergonomic risk assessments. Future work will aim to expand task diversity and integrate coupling assessment for full RNLE coverage. 

Fine-tuning large pretrained Transformer models can focus on either introducing a small number of new learnable parameters (parameter efficiency) or editing representations of a small number of tokens using lightweight modules (representation efficiency). While the pioneering method LoRA (Low-Rank Adaptation) inherently balances parameter, compute, and memory efficiency, many subsequent variants trade off compute and memory efficiency and/or performance to further reduce fine-tuning parameters. To address this limitation and unify parameter-efficient and representation-efficient fine-tuning, we propose Weight-Generative Fine-Tuning (WeGeFT, pronounced wee-gift), a novel approach that learns to generate fine-tuning weights directly from the pretrained weights. WeGeFT employs a simple low-rank formulation consisting of two linear layers, either shared across multiple layers of the pretrained model or individually learned for different layers. This design achieves multifaceted efficiency in parameters, representations, compute, and memory, while maintaining or exceeding the performance of LoRA and its variants. Extensive experiments on commonsense reasoning, arithmetic reasoning, instruction following, code generation, and visual recognition verify the effectiveness of our proposed WeGeFT. 

White-box targeted adversarial attacks reveal core vulnerabilities in Deep Neural Networks (DNNs), yet two key challenges persist: (i) How many target classes can be attacked simultaneously in a specified order, known as the ordered top-K attack problem (K ≥ 1)? (ii) How to compute the corresponding adversarial perturbations for a given benign image directly in the image space? We address both by showing that ordered top-K perturbations can be learned via iteratively optimizing linear combinations of the right singular vectors of the adversarial Jacobian (i.e., the logit-to-image Jacobian constrained by target ranking). These vectors span an orthogonal, informative subspace in the image domain. We introduce RisingAttacK, a novel Sequential Quadratic Programming (SQP)-based method that exploits this structure. We propose a holistic figure-of-merits (FoM) metric combining attack success rates (ASRs) and ℓp-norms (p = 1, 2, ∞). Extensive experiments on ImageNet-1k across six ordered top-K levels (K = 1, 5, 10, 15, 20, 25, 30) and four models (ResNet-50, DenseNet-121, ViTB, DEiT-B) show RisingAttacK consistently surpasses the state-of-the-art QuadAttacK. 

Existing score-based adversarial attacks mainly focus on crafting top-1 adversarial examples against classifiers with single-label classification. Their attack success rate and query efficiency are often less than satisfactory, particularly under small perturbation requirements; moreover, the vulnerability of classifiers with multilabel learning is yet to be studied. In this paper, we propose a comprehensive surrogate free score-based attack, named geometric score-based black-box attack (GSBAK), to craft adversarial examples in an aggressive top-K setting for both untargeted and targeted attacks, where the goal is to change the top-K predictions of the target classifier. We introduce novel gradient-based methods to find a good initial boundary point to attack. Our iterative method employs novel gradient estimation techniques, particularly effective in top-K setting, on the decision boundary to effectively exploit the geometry of the decision boundary. Additionally, GSBAK can be used to attack against classifiers with top-K multi-label learning. Extensive experimental results on ImageNet and PASCAL VOC datasets validate the effectiveness of GSBAK in crafting top-K adversarial examples. 

The adversarial vulnerability of Deep Neural Networks (DNNs) has been wellknown and widely concerned, often under the context of learning top-1 attacks (e.g., fooling a DNN to classify a cat image as dog). This paper shows that the concern is much more serious by learning significantly more aggressive ordered top-K clearbox 1 targeted attacks proposed in [Zhang and Wu, 2020]. We propose a novel and rigorous quadratic programming (QP) method of learning ordered top-K attacks with low computing cost, dubbed as QuadAttacK. Our QuadAttacK directly solves the QP to satisfy the attack constraint in the feature embedding space (i.e., the input space to the final linear classifier), which thus exploits the semantics of the feature embedding space (i.e., the principle of class coherence). With the optimized feature embedding vector perturbation, it then computes the adversarial perturbation in the data space via the vanilla one-step back-propagation. In experiments, the proposed QuadAttacK is tested in the ImageNet-1k classification using ResNet-50, DenseNet-121, and Vision Transformers (ViT-B and DEiT-S). It successfully pushes the boundary of successful ordered top-K attacks from K = 10 up to K = 20 at a cheap budget (1 × 60) and further improves attack success rates for K = 5 for all tested models, while retaining the performance for K = 1. 

Decision-based black-box attacks often necessitate a large number of queries to craft an adversarial example. Moreover, decision-based attacks based on querying boundary points in the estimated normal vector direction often suffer from inefficiency and convergence issues. In this paper, we propose a novel query-efficient \b curvature-aware \b geometric decision-based \b black-box \b attack (CGBA) that conducts boundary search along a semicircular path on a restricted 2D plane to ensure finding a boundary point successfully irrespective of the boundary curvature. While the proposed CGBA attack can work effectively for an arbitrary decision boundary, it is particularly efficient in exploiting the low curvature to craft high-quality adversarial examples, which is widely seen and experimentally verified in commonly used classifiers under non-targeted attacks. In contrast, the decision boundaries often exhibit higher curvature under targeted attacks. Thus, we develop a new query-efficient variant, CGBA-H, that is adapted for the targeted attack. In addition, we further design an algorithm to obtain a better initial boundary point at the expense of some extra queries, which considerably enhances the performance of the targeted attack. Extensive experiments are conducted to evaluate the performance of our proposed methods against some well-known classifiers on the ImageNet and CIFAR10 datasets, demonstrating the superiority of CGBA and CGBA-H over state-of-the-art non-targeted and targeted attacks, respectively. 

The main challenge of monocular 3D object detection is the accurate localization of 3D center. Motivated by a new and strong observation that this challenge can be remedied by a 3D-space local-grid search scheme in an ideal case, we propose a stage-wise approach, which combines the information flow from 2D-to-3D (3D bounding box proposal generation with a single 2D image) and 3D-to-2D (proposal verification by denoising with 3D-to-2D contexts) in a topdown manner. Specifically, we first obtain initial proposals from off-the-shelf backbone monocular 3D detectors. Then, we generate a 3D anchor space by local-grid sampling from the initial proposals. Finally, we perform 3D bounding box denoising at the 3D-to-2D proposal verification stage. To effectively learn discriminative features for denoising highly overlapped proposals, this paper presents a method of using the Perceiver I/O model [20] to fuse the 3D-to-2D geometric information and the 2D appearance information. With the encoded latent representation of a proposal, the verification head is implemented with a self-attention module. Our method, named as MonoXiver, is generic and can be easily adapted to any backbone monocular 3D detectors. Experimental results on the well-established KITTI dataset and the challenging large-scale Waymo dataset show that MonoXiver consistently achieves improvement with limited computation overhead. 

This study presents a mobile app that facilitates undergraduate students to learn data science through their own full body motions. Leveraging the built-in camera of a mobile device, the proposed app captures the user and feeds their images into an open-source computer-vision algorithm that localizes the key joint points of human body. As students can participate in the entire data collection process, the obtained motion data is context-rich and personally relevant to them. The app utilizes the collected motion data to explain various concepts and methods in data science under the context of human movements. The app also visualizes the geometric interpretation of data through various visual aids, such as interactive graphs and figures. In this study, we use principal component analysis, a commonly used dimensionality reduction method, as an example to demonstrate the proposed learning framework. Strategies to encompass other learning modules are also discussed for further improvement. 

Vision Transformers (ViTs) are built on the assumption of treating image patches as “visual tokens” and learn patch-to-patch attention. The patch embedding based tokenizer has a semantic gap with respect to its counterpart, the textual tokenizer. The patch-to-patch attention suffers from the quadratic complexity issue, and also makes it non-trivial to explain learned ViTs. To address these issues in ViT, this paper proposes to learn Patch-to-Cluster attention (PaCa) in ViT. Queries in our PaCa-ViT starts with patches, while keys and values are directly based on clustering (with a predefined small number of clusters). The clusters are learned end-to-end, leading to better tokenizers and inducing joint clustering-for-attention and attention-for-clustering for better and interpretable models. The quadratic complexity is relaxed to linear complexity. The proposed PaCa module is used in designing efficient and interpretable ViT backbones and semantic segmentation head networks. In experiments, the proposed methods are tested on ImageNet-1k image classification, MS-COCO object detection and instance segmentation and MIT-ADE20k semantic segmentation. Compared with the prior art, it obtains better performance in all the three benchmarks than the SWin [32] and the PVTs [47], [48] by significant margins in ImageNet-1k and MIT-ADE20k. It is also significantly more efficient than PVT models in MS-COCO and MIT-ADE20k due to the linear complexity. The learned clusters are semantically meaningful. Code and model checkpoints are available at https:/github.com/iVMCL/PaCaViT.