An optimized UNet segmentation model mapped to SoC FPGA hardware achieves 30 FPS jet flame analysis with 125x fewer parameters and no loss in Dice Score accuracy.
Journal of Computing in Civil Engineering 39, 04025068
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
A reconfigurable smart camera implementation for jet flames characterization based on an optimized segmentation model
An optimized UNet segmentation model mapped to SoC FPGA hardware achieves 30 FPS jet flame analysis with 125x fewer parameters and no loss in Dice Score accuracy.