The idea of semantic segmentation is to recognize and understand what is in an image at the pixel-level.
|Trend||Dataset||Best Method||Paper title||Paper||Code||Compare|
Recent progress has demonstrated that such meta-learning methods may exceed scalable human-invented architectures on image classification tasks.
#3 best model for Semantic Segmentation on Cityscapes (using extra training data)
The former networks are able to encode multi-scale contextual information by probing the incoming features with filters or pooling operations at multiple rates and multiple effective fields-of-view, while the latter networks can capture sharper object boundaries by gradually recovering the spatial information.
In this paper we describe a new mobile architecture, MobileNetV2, that improves the state of the art performance of mobile models on multiple tasks and benchmarks as well as across a spectrum of different model sizes.
#33 best model for Image Classification on ImageNet
To handle the problem of segmenting objects at multiple scales, we design modules which employ atrous convolution in cascade or in parallel to capture multi-scale context by adopting multiple atrous rates.
#3 best model for Semantic Segmentation on PASCAL VOC 2012
Our approach efficiently detects objects in an image while simultaneously generating a high-quality segmentation mask for each instance.
SOTA for Instance Segmentation on Cityscapes (using extra training data)
ASPP probes an incoming convolutional feature layer with filters at multiple sampling rates and effective fields-of-views, thus capturing objects as well as image context at multiple scales.
#7 best model for Semantic Segmentation on PASCAL Context
When we add our proposed global feature, and a technique for learning normalization parameters, accuracy increases consistently even over our improved versions of the baselines.
#12 best model for Semantic Segmentation on PASCAL Context
This is due to the very invariance properties that make DCNNs good for high level tasks.
SOTA for Scene Segmentation on SUN-RGBD