TY  - JOUR
T1  - Collaborative Inference Method Using Optimal Confidence Thresholds in Intelligent Surveillance Systems
AU  - Choi, Hyun-Ho 
JO  - The Journal of Korean Institute of Communications and Information Sciences
PY  - 2025
DA  - 2025/1/1
DO  - 10.7840/kics.2025.50.2.253
KW  - Cooperative inference
KW  - mobile edge computing
KW  - confidence threshold
KW  - surveillance.
AB  - This study proposes a novel collaborative inference method between edge device and edge server for intelligent surveillance services. In the proposed method, the device uses a small neural network (NN) and filters ambiguous input images by setting two confidence thresholds. These filtered images are then forwarded to the edge server for re-evaluation with a larger NN. Simulation results show that the proposed collaborative inference method, using optimal confidence thresholds, significantly reduces latency while meeting the required accuracy.