Light-Weight RefineNet for Real-Time Semantic Segmentation

8 Oct 2018  ·  Vladimir Nekrasov, Chunhua Shen, Ian Reid ·

We consider an important task of effective and efficient semantic image segmentation. In particular, we adapt a powerful semantic segmentation architecture, called RefineNet, into the more compact one, suitable even for tasks requiring real-time performance on high-resolution inputs... To this end, we identify computationally expensive blocks in the original setup, and propose two modifications aimed to decrease the number of parameters and floating point operations. By doing that, we achieve more than twofold model reduction, while keeping the performance levels almost intact. Our fastest model undergoes a significant speed-up boost from 20 FPS to 55 FPS on a generic GPU card on 512x512 inputs with solid 81.1% mean iou performance on the test set of PASCAL VOC, while our slowest model with 32 FPS (from original 17 FPS) shows 82.7% mean iou on the same dataset. Alternatively, we showcase that our approach is easily mixable with light-weight classification networks: we attain 79.2% mean iou on PASCAL VOC using a model that contains only 3.3M parameters and performs only 9.3B floating point operations. read more

PDF Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Semantic Segmentation NYU Depth v2 Light-Weight-RefineNet-152 Mean IoU 44.4% # 3
Real-Time Semantic Segmentation NYU Depth v2 Light-Weight-RefineNet-50 mIoU 41.7 # 7
Speed(ms/f) 20 # 4
Semantic Segmentation NYU Depth v2 Light-Weight-RefineNet-50 Mean IoU 41.7% # 7
Semantic Segmentation NYU Depth v2 Light-Weight-RefineNet-101 Mean IoU 43.6% # 4
Real-Time Semantic Segmentation NYU Depth v2 Light-Weight-RefineNet-152 mIoU 44.4 # 1
Speed(ms/f) 36 # 7
Real-Time Semantic Segmentation NYU Depth v2 Light-Weight-RefineNet-101 mIoU 43.6 # 2
Speed(ms/f) 27 # 5
Semantic Segmentation PASCAL VOC 2012 test Light-Weight-RefineNet-101 Mean IoU 82.0% # 33
Semantic Segmentation PASCAL VOC 2012 test Light-Weight-RefineNet-50 Mean IoU 81.1% # 34
Semantic Segmentation PASCAL VOC 2012 test Light-Weight-RefineNet-MobileNet-v2 Mean IoU 79.2% # 37
Semantic Segmentation PASCAL VOC 2012 test Light-Weight-RefineNet-152 Mean IoU 82.7% # 28

Methods


No methods listed for this paper. Add relevant methods here