TTC.PA.1623 : CONSTRUCTION SITE SAFETY
COMPLIANCE MONITORING
ENRICHED BY MULTI-CAMERA
WORKER RE-IDENTIFICATION
PROBLEM ADDRESSED
Construction worker re-identification and personal protective equipment detection across multiple cameras for continuous monitoring of site safety compliance over a large site area
APPLICATION AREAS
- Smart construction
workers’ safety & site analysis - Smart transportation
traffic flow analysis - Smart mobility
walkability evaluation - Aging city
elderly mobility improvement - Smart retail
behavioural analysis
TECHNOLOGY INNOVATIONS
A new loss function named Similarity Loss facilitates ReID robustness of different convolutional neural networks
A weighted-class model training strategy to enhance PPE classification given an imbalanced number of samples
A mechanism of incident logging to identify PPE safety incompliance of workers walking over a large site area
KEY IMPACTS & ADVANTAGES
- Accurate worker ReID
4%
improvement over baselines - PPE classification
13%
improvement over baselines
- Efficiently identity unsafe behaviour of workers on construction sites
- Continuously log down any incident of safety incompliance for devising safety measures
GET IN TOUCH
to receive a personalized consultation, discuss research or commercial collaboration opportunities, and more
info@autosafe.ai
+852 6538 6246