Learning Dense Convolutional Embeddings for Semantic Segmentation
This work addresses semantic segmentation for computer vision applications, but it is incremental as it builds on existing DCNN architectures with straightforward modifications.
The paper tackles the problem of semantic segmentation by proposing a deep convolutional neural network that learns pixel embeddings to distinguish same-region and boundary pixels, resulting in a systematic improvement in per-pixel classification accuracy when combined with an existing segmentation model.
This paper proposes a new deep convolutional neural network (DCNN) architecture that learns pixel embeddings, such that pairwise distances between the embeddings can be used to infer whether or not the pixels lie on the same region. That is, for any two pixels on the same object, the embeddings are trained to be similar; for any pair that straddles an object boundary, the embeddings are trained to be dissimilar. Experimental results show that when this embedding network is used in conjunction with a DCNN trained on semantic segmentation, there is a systematic improvement in per-pixel classification accuracy. Our contributions are integrated in the popular Caffe deep learning framework, and consist in straightforward modifications to convolution routines. As such, they can be exploited for any task involving convolution layers.