Science AI Journal
Computer Science · Research topic

Open research questions in Human Pose and Action Recognition

143 gap statements mined from Human Pose and Action Recognition papers in our 4.5M-paper local library, which holds 1,317 papers on the topic — drawn mostly from each paper's own stated research gap, future-work, challenge and limitation notes, and its abstract. The ones listed below are a selection still marked open; each names the study that raised it, with a DOI link where the paper has one.

Choosing where to publish on Human Pose and Action Recognition? See the ranked Computer Science journals.

What the literature leaves open

  • Future studies will focus on lightweight ViTs, multimodal data fusion, and self-supervised learning to facilitate the implementation of these approaches in various areas, including smart surveillance, healthcare monitoring, AR/VR interfaces, and human-robot collaboration in smart environments.

    ViT-HAR: Vision Transformer-Based Human Activity Recognition in Cluttered Environments · 2026 · DOI

  • To address this limitation, future work will focus on expanding the dataset to include a broader spectrum of postures and gestures and on exploring more advanced learning strategies to improve robustness and generalization.

    Human pose estimation based engagement detection in E-learning: a hybrid approach · 2026 · DOI

  • There is a lack of study on using real-time surveillance cameras to identify armed individuals. Current systems rely on human supervision, which can be ineffective.

    Real-Time Detection of Armed Individuals in Video Surveillance Using Machine Learning Techniques · 2026 · DOI

  • Real-time human action recognition deployment in Internet of Things environments remains limited by latency, bandwidth, and the tradeoff between model size and performance.

    Skeleton-Based Human Action Recognition Using Multitype Input Data on Edge Devices in Internet of Things Systems · 2026 · DOI

  • The risk of suffering from injuries due to lack of professional training. The need for real-time exercise form monitoring and corrective capabilities. The requirement for automated systems that can analyze human body movements.

    AI Fitness Coach : Real-Time Exercise Posture Correction Using MediaPipe Pose Estimation and Voice Guidance · 2026 · DOI

  • The need for innovative solutions to master yoga poses, particularly for beginners lacking access to experienced instructors. The lack of lightweight and efficient models that can provide accurate pose estimation while conserving computational resources.

    Yoga Pose Estimation and Prediction · 2026 · DOI

  • Using the current design, several methods may be used to further enhance the system, including: Real-Time Correction and Feedback: Customers can get real-time feedback by integrating posture correction assistance with biomechanical angle analysis. Develop lightweight transformers or quantized models that may be used on wearables and mobile devices to optimize edge deployment.

    Hybrid vision–IMU deep learning framework with graph convolutional networks and attention for personalized yoga posture identification · 2026 · DOI

  • The need for trustworthy AI systems to support caregivers in kindergartens. The lack of effective monitoring systems for ensuring child safety.

    Real-Time Monitoring of Kindergarten Safety Using YOLO-11-Based Detection of Children and Adults · 2026 · DOI

  • Future research directions include supporting more complex interaction dynamics involving multiple objects and multi-agent settings. Future research directions include addressing the limitations of the simulator, such as modeling social or collaborative scenarios.

    EgoInteract: Synthetic Egocentric Videos Generation for Interaction Understanding and Anticipation · 2026

  • The research gap is the lack of large-scale egocentric video datasets with dense spatial and temporal annotations. The research gap is the limited coverage of interaction dynamics in existing datasets, hindering the generalization of models across tasks and scenarios.

    EgoInteract: Synthetic Egocentric Videos Generation for Interaction Understanding and Anticipation · 2026

  • Joint object-hand modeling and end-to-end training are natural next steps. Applying the quality-aware flow-matching framework to other domains with similar challenges.

    StableHand: Quality-Aware Flow Matching for World-Space Dual-Hand Motion Estimation from Egocentric Video · 2026

  • The system requires a custom exoskeleton backpack and Apple Vision Pro for markerless hand tracking. The dataset is limited to 10K teleoperated episodes and 2.92M frames.

    Dexora: Open-source VLA for High-DoF Bimanual Dexterity · 2026

  • Future work should examine model compression techniques, efficient quantization, lightweight model structures, and energy-aware training procedures to ensure the hybrid model remains practical for ubiquitous deployment.

    KAN-MLP-Mixer: A comprehensive investigation of the usage of Kolmogorov-Arnold Networks (KANs) for improving IMU-based Human Activity Recognition · 2026

  • The model is not evaluated for real-time pose tracking or monocular 3D human pose estimation. The dataset used is limited to MS COCO 2017.

    A Lightweight Human Pose Estimation Algorithm Based on Improved YOLO11-Pose · 2026 · DOI

  • The literature on HAR is vast, but researchers face challenges in selecting optimal architectures and models. There is a need for a comprehensive review of HAR methods to identify the best performing models.

    Human Action Recognition Based on Computer Vision:A Comprehensive Review (2020-2025) · 2026 · DOI

  • Current supervised learning methods are bound by the availability of labeled data, restricting them to laboratory environments. The model may not generalize to real-world or previously unseen movements. The dataset is limited to 10 participants. The model is restricted to a sagittal-plane lower limb model.

    SSPINNpose: A Self-Supervised PINN for Inertial Pose and Dynamics Estimation · 2026 · DOI

  • Current real-time methods rely on supervised learning, which requires ground truth datasets measured with laboratory measurement systems. These systems introduce measurement and processing errors and often fail to generalize to real-world movements. There is a need for a method that can accurately estimate human movement dynamics without requiring labeled training data.

    SSPINNpose: A Self-Supervised PINN for Inertial Pose and Dynamics Estimation · 2026 · DOI

  • The system requires users to complete their workouts while being filmed by cameras. The system does not need advanced GPUs but uses Python computer vision libraries. Future enhancements will enable the system to support more advanced exercises.

    AI Fitness Coach : Real-Time Exercise Posture Correction Using MediaPipe Pose Estimation and Voice Guidance · 2026 · DOI

  • Limited accuracy of standard head pose benchmarks for head acceleration event conditions. Difficulty in estimating head pose during large, multi-axis rotations.

    Deep learning approaches for head pose estimation in sports impacts · 2026 · DOI

  • The dataset was limited to 10 participants. The study did not evaluate the models on other types of head acceleration events.

    Deep learning approaches for head pose estimation in sports impacts · 2026 · DOI

  • The difficulty of obtaining large amounts of labeled data for supervised skeleton-based action recognition. The limitation of simple late fusion in enhancing performance while maintaining model efficiency.

    Multimodal Skeleton-based Action Representation Learning via Decomposition and Composition · 2026 · DOI

  • Existing methods uniformly condition on noisy hand motion observations without accounting for their per-frame reliability. The egocentric capture setting poses significant challenges for recovering the world space 4D motion signal.

    StableHand: Quality-Aware Flow Matching for World-Space Dual-Hand Motion Estimation from Egocentric Video · 2026

  • Extending Dexora to more complex tasks and domains. Improving the robustness and generalization of the system. Exploring the application of Dexora to various fields, including robotics, manufacturing, and healthcare.

    Dexora: Open-source VLA for High-DoF Bimanual Dexterity · 2026

  • The absence of posture analysis in existing fitness applications - The lack of real-time corrective feedback in existing fitness applications - The dependence on wearable devices in existing fitness…

    AI-Powered Real-Time Fitness Tracking and Posture Detection System · 2026 · DOI

  • Extending the temporal context to 15-30 seconds to improve recall for subtle categories. Incorporating trajectory history to improve detection accuracy. Developing more accurate suspicious behavior detection systems using the proposed SUSP-POSE dataset.

    Construction and Benchmarking of a Dataset for Human Pose Estimation of Suspicious Behavior in Outdoor Environments · 2026 · DOI

Most-cited papers in Human Pose and Action Recognition

Most recent work

Find a gap in your own Human Pose and Action Recognition sub-topic

This page shows what the Human Pose and Action Recognition literature already flags as unresolved. To narrow it to your specific question, search the Research Gap Finder: the search is free with a free account and lists the papers closest to your topic first. Unlocking that topic (50 credits, charged once) fills the comparison table from our 4.5M-paper local library and writes the gaps from its rows.

Open the Research Gap Finder →

Related topics in Computer Science

Online Learning and Analytics
887 open questions
AI in Service Interactions
805 open questions
Quantum Computing Algorithms and Architecture
632 open questions
Virtual Reality Applications and Impacts
610 open questions
Teaching and Learning Programming
576 open questions
Digital literacy in education
522 open questions
Computational Drug Discovery Methods
505 open questions
Topic Modeling
482 open questions
Open questions raised by papers listed on this page include: (1) To address this limitation, future work will focus on expanding the dataset to include a broader spectrum of postures and gestures and on exploring more advanced learning strategies to improve robustness and generalization (“Human pose estimation based engagement detection in E-learning: a hybrid…”, 2026); (2) There is a lack of study on using real-time surveillance cameras to identify armed individuals (“Real-Time Detection of Armed Individuals in Video Surveillance Using Machine…”, 2026); (3) Real-time human action recognition deployment in Internet of Things environments remains limited by latency, bandwidth, and the tradeoff between model size and performance (“Skeleton-Based Human Action Recognition Using Multitype Input Data on Edge…”, 2026); (4) The risk of suffering from injuries due to lack of professional training (“AI Fitness Coach : Real-Time Exercise Posture Correction Using MediaPipe Pose…”, 2026); (5) The lack of lightweight and efficient models that can provide accurate pose estimation while conserving computational resources (“Yoga Pose Estimation and Prediction”, 2026). 143 gap statements have been mined from Human Pose and Action Recognition papers in our 4.5M-paper local library, which holds 1,317 papers on the topic; the gaps come from whichever of those papers state one. They are mostly the research gaps the authors state and the papers' abstracts, plus future-work, limitations and challenges passages. The ones listed below are a selection still marked open; each names the study that raised it, with a DOI link where the paper has one, so you can read the original claim in context.

Tools for your next paper

Gaps & Literature SearchFind open questions in any field — 120,000+ open gaps mined from our own library.Pre-CheckIs your paper ready? Acceptance odds for top, mid and open/emerging journals, in seconds.Journal FinderA ranked shortlist of target journals from a 17,500-venue index.Duplicate Publication CheckerCheck for prior publication across 7 sources and our local library.Citation GeneratorFreeA DOI or title → APA, MLA, Vancouver, Chicago, BibTeX, RIS. No signup.Graphical Abstract MakerFreeTurn your findings into a colorblind-safe graphical abstract.

How Pre-Check worksHow AI Review works

Compare the category — Honest roundups of the AI research tools, ours listed alongside the alternatives.

Command palette

Jump anywhere, run any action.