Hierarchical Deep Learning Architectures for Multiclass Violation Detection in Urban Surveillance Systems Authors
Bibliographic record
Abstract
Urban surveillance systems face the challenge of recognizing a wide spectrum of violations in real time under diverse environmental conditions. This study proposes a hierarchical deep learning architecture that integrates convolutional neural networks (CNNs) for spatial feature extraction and bidirectional recurrent neural networks (RNNs) for temporal sequence modeling. A large-scale dataset was constructed, comprising 120,000 labeled video segments across 172 violation categories, each annotated by three independent experts with an inter-annotator agreement of κ = 0.87. To ensure rigorous evaluation, the dataset was split into training (80%), validation (10%), and testing (10%) sets, and control baselines including CNN-only, RNN-only, and flat CNN-RNN models were established. Experimental results demonstrate that the proposed framework achieved a macro-averaged F1-score of 0.961, outperforming the baseline models by margins ranging from 12% to 17%. Robustness tests under illumination shifts, occlusion, and camera angle variations showed less than 3% degradation in accuracy, compared with more than 10% performance loss in baseline models. Furthermore, hierarchical classification reduced misclassification among semantically similar events, such as "illegal parking" and "lane obstruction," which were frequently confused in flat architectures. The scalability of the framework was confirmed by its ability to handle over 170 categories without performance decline, a capability rarely reported in prior studies. These findings indicate that hierarchical spatial-temporal modeling not only improves classification accuracy but also enhances robustness and scalability, making it a promising solution for intelligent urban surveillance. The proposed architecture offers practical significance for real-world deployment and provides methodological insights for the design of next-generation smart surveillance systems.
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.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 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 itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".