Stella Yu : Publications

Reconstructive Sparse Code Transfer for Contour Detection and Semantic Labeling
Michael Maire and Stella X. Yu and Pietro Perona
Asian Conference on Computer Vision, Singapore, 1-5 November 2014
Paper | Slides


We frame the task of predicting a semantic labeling as a sparse reconstruction procedure that applies a target-specific learned transfer function to a generic deep sparse code representation of an image. This strategy partitions training into two distinct stages. First, in an unsupervised manner, we learn a set of dictionaries optimized for sparse coding of image patches. These generic dictionaries minimize error with respect to representing image appearance and are independent of any particular target task. We train a multilayer representation via recursive sparse dictionary learning on pooled codes output by earlier layers. Second, we encode all training images with the generic dictionaries and learn a transfer function that optimizes reconstruction of patches extracted from annotated ground-truth given the sparse codes of their corresponding image patches. At test time, we encode a novel image using the generic dictionaries and then reconstruct using the transfer function. The output reconstruction is a semantic labeling of the test image.

Applying this strategy to the task of contour detection, we demonstrate performance competitive with state-of-the-art systems. Unlike almost all prior work, our approach obviates the need for any form of hand-designed features or filters. Our model is entirely learned from image and ground-truth patches, with only patch sizes, dictionary sizes and sparsity levels, and depth of the network as chosen parameters. To illustrate the general applicability of our approach, we also show initial results on the task of semantic part labeling of human faces.

The effectiveness of our data-driven approach opens new avenues for research on deep sparse representations. Our classifiers utilize this representation in a novel manner. Rather than acting on nodes in the deepest layer, they attach to nodes along a slice through multiple layers of the network in order to make predictions about local patches. Our flexible combination of a generatively learned sparse representation with discriminatively trained transfer classifiers extends the notion of sparse reconstruction to encompass arbitrary semantic labeling tasks.

image segmentation, multiscale, sparse coding, dictionary learning