A Hybrid Edge-Cloud Smart Door Surveillance System with Real-Time Risk Assessment and Secure Access Control
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
This paper presents a hybrid edge-cloud smart door surveillance system designed to enhance residential security through real-time video analytics, anomaly detection, and dynamic access control. A local webcam captures live video streams processed at the edge by a Flask-Based server, integrating YOLOv8 object detection, HMM-based behavior anomaly detection, and facial recognition for identity verification. To guarantee uninterrupted operation without network connectivity, we also implement a fully offline face-recognition mode on the ESP32-CAM enabling the door to unlock locally under 100 ms. In parallel, a retrainable cloud-based Azure Custom Vision model detected object identities, assessed risk levels dynamically triggering local actions such as door control and real-time email alerts with contextual evidence. When risk levels exceed predefined thresholds, the system autonomously captures critical frames and sends alert or action-required notifications with attached visual evidence, using risk-specific visual theming. To address edge computing limitations in complex cases, recorded videos are analyzed by AWS Rekognition for comprehensive label detection. A responsive web interface provides live monitoring, event logging, OTP-based visitor access, and remote door operation, enhanced by dynamic UI feedback based on threat severity. This work demonstrates an integrated, scalable, and cost-efficient approach to edge-assisted smart security systems, laying a foundation for future improvements in multi-camera coordination ensuring the embedded firmware remains lightweight and easily updatable.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.000 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it