DNN-SAM accepted at IEEE RTAS 2022

Our paper “DNN-SAM: Split-and-Merge DNN Execution for Real-Time Object Detection” was accepted at the 28th IEEE Real-Time and Embedded Technology and Applications Symposium.

DNN-SAM pipeline splitting an image into mandatory and optional networks, scheduling their subtasks, and merging their results — DNN-SAM splits, independently schedules, and merges mandatory and optional DNN subtasks.

DNN-SAM separates object-detection inference into mandatory and optional subtasks, schedules them according to criticality, and merges their outputs. This design improves responsiveness and accuracy in safety-critical image regions while meeting timing constraints.

The paper is available from IEEE Xplore.

Source: RTCL@DGIST announcement