Open research questions in Face recognition and analysis
30 unresolved questions extracted from the limitations and future-work sections of 420 Face recognition and analysis papers in our library. Each links back to the study that raised it.
What the literature leaves open
includes Arc Face and Mobile FaceNet optimisation, anti-spoofing mechanisms, mobile application support, cloud synchronisation, ERP integration, federated learning, and multi-camera deployment. Emotion analytics, classroom engagement tracking, and Edge TPU acceleration can intelligent educational improve monitoring. VIII. CONCLUSION AND FUTURE SCOPE This paper presented a Smart Attendance System integrating Raspberry Pi 4 edge computing, real-time facial recognition, and a Flask-based web management platform. The system demonstrated >92% recognition accuracy under standard conditions, complete proxy prevention, and a deployment cost of INR 11,097. The multi-device REST API enables campus- wide deployment with centralized management. Future enhancements include: (1) ArcFace/FaceNet models with TFLite quantization for improved accuracy under adverse conditions, (2) a mobile application for Android/iOS, (3) liveness detection (4) ERP/LMS integration, and facial expression-based engagement analytics. to prevent spoofing, (5) ACKNOWLEDGMENT The authors express sincere gratitude to the Department of Electronics and Telecommunication Engineering, PES's College of Engineering Phaltan, for providing laboratory infrastructure. The authors also acknowledge the open-source communities behind Python, OpenCV, dlib, Flask, and PostgreSQL. REFERENCES 1. Zahid SMKB, Nishat MRH, Hasib A, Hasan MR, Ashiqussalehin M, Sajib MSH. Real-time multi-modal embedded vision framework for object detection, facial emotion recognition and biometric identification on low- power edge platforms. arXiv preprint arXiv:2601.11970. 2026 Jan. 2. Turpo Benique CO. School attendance control system based on RFID technology with Raspberry Pi and Arduino. arXiv preprint arXiv:2507.14191. 2025 Jul. 3. Ainebyona K, Oguti AM, Walusimbi J, Kobusingye R. Integrating attendance tracking and emotion detection for enhanced student engagement in smart classrooms. Smart Classroom Systems Journal. 2025. 4. Yadav U. A web-based facial recognition attendance system. Bachelor’s thesis. LAB University of Applied Sciences, Finland. 2025. 5. Abderraouf T, Wassim AA, Larabi S. An embedded for attendance monitoring. arXiv intelligent system preprint arXiv:2406.13694. 2024 Jun. 6. Azmi F, Saleh A, Ridwan A. Smart management attendance system with facial recognition using computer vision techniques on Raspberry Pi. Int J Innov Res Comput Sci Technol. 2023;11(1):38–44. 7. Deng J, Guo J, Xue N, Zafeiriou S. ArcFace: additive angular margin loss for deep face recognition. IEEE Trans Pattern Anal Mach Intell. 2022;44(10):5962–5979. 8. Singh V, Mehta D. Multi-modal classroom attendance with emotion detection. In: Proc IEEE ICCCT. 2022. p. 1–6. 9. Kumar S, Sharma A, Verma R. Edge computing for real- time intelligent monitoring. IEEE Internet of Things Journal. 2021;8(5):3421–3432. 616 © 2026 Prof. Amarsinha Ashokrao Ranaware, Mayuresh Dharmadhikari, Priti Kumbhar, Sejal Galage, Mayuri Kadam. This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International License (CC BY NC ND).https://creativecommons.org/licenses/by/4.0/ Int. Jr. of Contemp. Res. in Multi. PEER-REVIEWED JOURNAL Volume 5 Issue 3 [May-Jun] Year 2026 10. Yadav R, Sharma P. Smart classroom management with attendance automation. Int J Adv Comput Sci Appl. 2021;12(3):145–153. 11. Mishra A, Tiwari K, Gupta B. Deep learning based Sciences. system. Applied attendance automated 2021;11(8):3606. 12. Patel AK, Shah M, Joshi N. Comparative analysis of attendance management systems. Journal of Engineering and Technology. 2020;9(2):78–89. Creative Commons (CC) License This article is an open-access article distributed under the terms and conditions of the Creative Commons Attribution–Non-Commercial–No Derivatives 4.0 International (CC BY-NC-ND 4.0) license. This license permits sharing and redistribution of the article in any medium or format for non-commercial purposes only, provided that appropriate credit is given to the original author(s) and source. No modifications, adaptations, or derivative works are permitted under this license.
Smart Attendance System: A Raspberry Pi-Powered Face Recognition and Automated Attendance Management System Using Webcam Integration · 2026 · DOI1) Deepfake detection continues to develop thanks to improvements in artificial intelligence and security technologies. In particular, nowadays, researchers do not Volume 14 Issue 5, May 2026 www.ijser.in Licensed Under Creative Commons Attribution CC BY Paper ID: SE26508232113DOI: https://dx.doi.org/10.70729/SE2650823211327 of 28 International Journal of Scientific Engineering and Research (IJSER) ISSN (Online): 2347-3878 SJIF (2025): 8.036 on the generalization problem, showing that many detection models struggle when tested on unseen datasets. (ScienceDirect) Bharati, N. (2025). Explainable deepfake detection framework. This research introduces explainable AI (XAI) techniques in deepfake detection, making results more interpretable for forensic and legal applications. (ScienceDirect) Khan, A. A. et al. (2025). Survey on multimediaenabled deepfake detection. The paper discusses multimodal detection systems combining video, audio, and facial cues to improve accuracy and robustness. (Springer) Qureshi, S. M. et al. (2024). Survey of digital forensic methods for multimodal deepfake detection. This study explores detection across multiple media types and highlights challenges in detecting deepfake content on social media platforms. (PMC) Gupta, G. (2023). Deepfake detection using multimodal and machine learning approaches. The research into datasets, benchmarks, and provides machine learning techniques used in deepfake detection systems. (MDPI) insights Wang, T. (2022). Deepfake detection: A reliabilityfocused survey. This paper emphasizes challenges such as robustness, transferability, and interpretability in detection models. (arXiv) ResearchGate Survey (2025). A survey on deepfake video detection. This review highlights the current state of deepfake detection and stresses the need for better real-world applicability and robustness. (ResearchGate) Banerjee, S. (2025). Survey on deepfake detection technologies. The paper examines both traditional and modern detection methods, including biometric and AIbased techniques. (ResearchGate) IJRCSEIT (2026). Deepfake detection through deep learning. This study compares different neural network architectures such as CNNs, RNNs, and transformers in terms of performance and efficiency. (IJSRCSEIT) Singh, S. (2025). Integrative review of deepfake detection and multimedia forensics. The authors discuss the social and cybersecurity implications of deepfakes and suggest interdisciplinary solutions. (PMC) limit themselves to studying the number of blinks; instead, they try to develop a system capable of recognizing changes and adapting to them. it is important 2) Multimodal biometric techniques do not imply only facial features; to analyze audio characteristics in addition. Today's security technologies use not only face recognition and verification but also fingerprint, voice, and multifactor authentications. 3) Thanks to the development of real-time detection technology, it becomes possible to prevent attempts to post any deepfake videos online instantly. In addition, blockchain technology should be regarded as promising within this area. 4) Finally, the role of datasets increases since face detection models prepare fight with deepfakes. the Simultaneously, detection tools are improving, and privacy laws are introduced.
In this survey, we provide a comprehensive overview of deep learning-based face video restoration techniques. We systematically categorize existing FVR methods along three primary dimensions: network architecture, temporal modeling strategies, and facial detail enhancement strategies. The discussion also encompasses commonly used datasets and evaluation metrics pertinent to FVR research. This survey is intended to serve as a reference, offering researchers insights into the key methodologies and advancements in FVR field. We hope that this survey could aid researchers in developing more effective techniques and contribute to the continued progress. Despite the significant progress achieved by previous research in FVR, several challenges remain to be addressed. We highlight key limitations in the current approaches and propose potential future research directions to further advance the field in the following. Robustness to real-world degradations. A primary challenge in FVR is enhancing model robustness to the complex and diverse degradations present in realworld videos. Current methods, frequently trained on synthetic or limited real-world datasets, may not generalize well to unseen degradation types, such as severe blur, extreme noise, and intricate compression artifacts. Future work could focus on more generalizable and interpretable blind modeling frameworks, as well as integrating degradation-aware mechanisms into temporal modeling for better spatio-temporal consistency. Additionally, self-supervised and unsupervised learning paradigms hold great potential for leveraging large-scale unlabeled C. Wang et al.
Furthermore, hardware as lowresolution cameras or limited processing power can affect system efficiency. High-quality input data and adequate computational resources are necessary to achieve optimal performance. such © 2026 The Author(s). Published by IJCOPE Journal. Website: https://ijcope.org/ 6 International Journal of Creative and Open Research in Engineering and Management ISSN: 3108-1754 (Online) Volume 02 Issue 05 May-2026 | Impact Factor: 3.5 related Overall, these challenges highlight the need for continuous improvement in system design. Addressing issues illumination, occlusion, pose variation, and computational efficiency can further enhance the robustness and reliability of the proposed attendance system.
AI-generated face detection fairness evaluation is currently limited to facial imagery. Future work should extend fairness competitions to broader visual contexts and explicitly measure performance trade-offs between faces and other image types to create more challenging and realistic evaluation scenarios.
The role and permissibility of post-processing techniques in fair deepfake detection remains undefined. A systematic examination is needed to determine whether post-processing should be allowed in fairness-focused competitions and how to establish equitable comparison standards when different teams employ different post-processing strategies.
Evaluation metrics for fairness in AI-generated face detection need refinement to explicitly penalize trivial solutions that improve fairness metrics at the expense of detection utility. Current metrics do not sufficiently incentivize practically deployable methods that maintain balanced performance across both fairness and accuracy objectives.
The drag-and-drop sketch construction interface is described qualitatively as requiring no artistic expertise, but no user study data, error rates for untrained users, or comparison of sketch quality produced by forensic artists versus novice users is provided to validate usability claims.
No ablation studies or comparative validation are provided to determine which deep learning architecture components (feature encoders, generative adversarial networks, domain alignment) contribute most to the reported 90% accuracy in sketch-photo synthesis and recognition.
The system claims social media platform integration as a future capability, but does not address how sketch-to-photo matching will perform on unconstrained, low-quality social media images compared to controlled forensic databases, or how privacy and consent issues will be handled when matching sketches against social media profiles.
The paper mentions integration with 3D face mapping and real-time CCTV surveillance matching as future features, but provides no methodological details on how sketch-based recognition will scale to large-scale surveillance databases or how the matching algorithm will handle temporal variations in appearance (aging, facial hair, expressions) across extended surveillance periods.
The deep learning recognition model achieved over 90% accuracy on unspecified datasets, but the paper does not report performance across diverse sketch types (e.g., composite sketches vs. hand-drawn sketches, varying sketch quality levels, or different demographic populations). Testing the sketch-photo matching accuracy across heterogeneous sketch construction methods and quality variations is needed.
ITA categorisation has been rather uncritically endorsed in computer vision research despite prioritising whiteness and erroneously regenerating race-based classifications disguised as neutral skin-type analysis.
Dermatological skin tone classifications are biased towards lighter skin types, with the original ITA and Fitzpatrick classifications having four categories for light skin types and only two for darker ones.
Therefore, new methods of detection should be explored as a matter of urgency since the latest 'off the shelf' AI tools can now generate face images of real people that are essentially undetectable as synthetic to most human observers.
In Experiment 3 (UK and Japan; n = 407), passers-by failed to notice that a live confederate was wearing a hyper-realistic mask and showed limited evidence of covert detection, even at close viewing distance (5 vs.
There is a need to address subjectivity of annotations by providing detailed documentation of how subjectivity has been taken into account in research on skin type classification.
Skin type annotation and colour categorisation is inherently subjective, with crowdsourced workers not systematically categorising pictures into different skin types in the same way.
The paper lacks discussion on privacy implications, data storage of video streams and alerts, and compliance with privacy regulations in residential surveillance.
Real-Time Edge-Based Burglary Detection and Automated Alerting Using Deep Learning Framework · 2026 · DOIHowever, the modeling accuracy and robustness of existing technologies in dealing with weak texture areas and complex lighting conditions are insufficient, which limits their practical application in production.
The issue of face recognition is not completely new on a worldwide scale, however, researchers present conflicting results and raise new questions about this phenomenon.
Most-cited papers in Face recognition and analysis
- YOLO-FaceV2: A scale and occlusion aware face detector · Pattern Recognition · 2024 · 372 citations
- Deep Learning-Based Face Mask Recognition in Real-Time Photos and Videos · African Journal of Biomedical Research · 2024 · 128 citations
- GauHuman: Articulated Gaussian Splatting from Monocular Human Videos · 2024 · 125 citations
- Diffused Heads: Diffusion Models Beat GANs on Talking-Face Generation · 2024 · 122 citations
- Surgical face masks impair human face matching performance for familiar and unfamiliar faces · Cognitive Research Principles and Implications · 2020 · 108 citations
- A Benchmark of Facial Recognition Pipelines and Co-Usability Performances of Modules · Bilişim Teknolojileri Dergisi · 2024 · 96 citations
- Expressive 3D Facial Animation Generation Based on Local-to-Global Latent Diffusion · IEEE Transactions on Visualization and Computer Graphics · 2024 · 87 citations
- TalkingStyle: Personalized Speech-Driven 3D Facial Animation With Style Preservation · IEEE Transactions on Visualization and Computer Graphics · 2024 · 80 citations
- DiffPoseTalk: Speech-Driven Stylistic 3D Facial Animation and Head Pose Generation via Diffusion Models · ACM Transactions on Graphics · 2024 · 76 citations
- Sensitive loss: Improving accuracy and fairness of face representations with discrimination-aware deep learning · Artificial Intelligence · 2022 · 69 citations
Most recent work
- Is this real? Susceptibility to deepfakes in machines and humans · Cognitive Research Principles and Implications · 2026
- Real-Time Edge-Based Burglary Detection and Automated Alerting Using Deep Learning Framework · International Research Journal on Advanced Engineering and Management (IRJAEM) · 2026
- Facial analysis AI as social pseudotechnology · European Journal for Philosophy of Science · 2026
- Criminal Face Sketch Recognition and Construction · International Journal for Research in Applied Science and Engineering Technology · 2026
- DeepFake Face Detection using Machine Learning · International Research Journal of Modernization in Engineering Technology & Science · 2026
- Supervised Neural Style Transfer as an Augmentation Technique for Facial Landmark Detection · International Journal of Computer Vision · 2026
- Face Identification from Obfuscated Images in Deep Learning using Feature Compensation · International Scientific Journal of Engineering and Management · 2026
- The Competition of Fairness in AI-generated Face Detection: Methods and Results · Machine Intelligence Research · 2026
- Dual-Module Uncalibrated Photometric Stereo with Illumination Estimation and Robust Normal Regression · Engineering Research Express · 2026
- Smart Attendance System Using Artificial Intelligence Based Face Recognition · International Journal of Creative and Open Research in Engineering and Management · 2026
Find a gap in your own Face recognition and analysis sub-topic
This page shows what the Face recognition and analysis literature already flags as unresolved. To narrow it to your specific question, run the guided finder — it searches the gap library on demand and checks candidates against 250M+ OpenAlex works.
Open the Research Gap Finder →