AI-Based Real-Time Fight Detection Through CCTV Cameras
Bibliographic record
Abstract
Detecting violent behavior, particularly physical fights, in real-time video streams is a critical task to ensure public safety in environments such as schools, prisons, and public surveillance systems. Currently, most surveillance systems rely on manual monitoring through CCTV cameras, which is inherently limited due to operator fatigue, labor intensiveness, delayed reaction times, subjectivity, and poor scalability in multi-camera setups. These limitations often result in missed or delayed detection of violent incidents. Although several research efforts have proposed automated human activity detection, such as fights, many of these methods suffer from low accuracy or are impractical for real-time deployment due to computational inefficiencies. In this paper, we propose an efficient and accurate neural network-based system for real-time fight detection. Specifically, we employ a transfer learning approach that combines a pre-trained InceptionV3 model for spatial feature extraction with the temporal sequence modeling capabilities of recurrent neural networks (RNNs), including LSTM and GRU variants. Our system analyzes video streams frame-by-frame in real-time, using the InceptionV3 model to extract spatial features from each frame. These features are then passed on to the RNN to learn temporal patterns across frames and predict violent activity. Based on the number of frames classified as violent within a specified time window, the system determines whether a fight is occurring and triggers an alert accordingly. Unlike traditional heavy-weight models, our approach emphasizes computational efficiency without compromising detection performance. Experimental results demonstrate that our system achieves over$\mathbf{9 0 \%}$in accuracy, precision, recall, and F1-score, while maintaining real-time processing capabilities.
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 | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
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".