# High-performance Semantic Segmentation Using Very Deep Fully Convolutional Networks

arXiv: https://arxiv.org/abs/1604.04339 · HTML: https://mlsanity.com/paper/1604.04339/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 15 Apr 2016.

## Built on by (most cited first, 9 of 9)

- [TransVOD: End-to-End Video Object Detection with Spatial-Temporal Transformers](https://mlsanity.com/paper/2201.05047/) (2022)
- [Semantic Segmentation with Labeling Uncertainty and Class Imbalance](https://mlsanity.com/paper/2102.04566/) (2021)
- [One shot PACS: Patient specific Anatomic Context and Shape prior aware recurrent registration-segmentation of longitudinal thoracic cone beam CTs](https://mlsanity.com/paper/2201.11000/) (2022)
- [SHREC 2022: pothole and crack detection in the road pavement using images and RGB-D data](https://mlsanity.com/paper/2205.13326/) (2022)
- [Category Feature Transformer for Semantic Segmentation](https://mlsanity.com/paper/2308.05581/) (2023)
- [Sequential Ensembling for Semantic Segmentation](https://mlsanity.com/paper/2210.05387/) (2022)
- [Attention-based fusion of semantic boundary and non-boundary information to improve semantic segmentation](https://mlsanity.com/paper/2108.02840/) (2021)
- [Local-to-Global Information Communication for Real-Time Semantic Segmentation Network Search](https://mlsanity.com/paper/2302.08481/) (2023)
- [GPU Memory Usage Optimization for Backward Propagation in Deep Network Training](https://mlsanity.com/paper/2502.12499/) (2025)

## Abstract (arXiv)

We propose a method for high-performance semantic image segmentation (or semantic pixel labelling) based on very deep residual networks, which achieves the state-of-the-art performance. A few design factors are carefully considered to this end. We make the following contributions. (i) First, we evaluate different variations of a fully convolutional residual network so as to find the best configuration, including the number of layers, the resolution of feature maps, and the size of field-of-view. Our experiments show that further enlarging the field-of-view and increasing the resolution of feature maps are typically beneficial, which however inevitably leads to a higher demand for GPU memories. To walk around the limitation, we propose a new method to simulate a high resolution network with a low resolution network, which can be applied during training and/or testing. (ii) Second, we propose an online bootstrapping method for training. We demonstrate that online bootstrapping is critically important for achieving good accuracy. (iii) Third we apply the traditional dropout to some of the residual blocks, which further improves the performance. (iv) Finally, our method achieves the currently best mean intersection-over-union 78.3% on the PASCAL VOC 2012 dataset, as well as on the recent dataset Cityscapes.