Exploring Semantic Segmentation on the DCT Representation

TL;DR

This paper investigates performing semantic segmentation directly on JPEG DCT coefficients, demonstrating comparable accuracy to RGB models with fewer coefficients and robustness to quantization errors, marking a novel approach in the field.

Contribution

It introduces the first method for semantic segmentation on DCT representations, optimizing coefficient selection for efficiency and robustness.

Findings

Achieves near-RGB accuracy using only 36% of DCT coefficients.

Demonstrates robustness of segmentation performance under quantization errors.

First to explore semantic segmentation directly on JPEG DCT data.

Abstract

Typical convolutional networks are trained and conducted on RGB images. However, images are often compressed for memory savings and efficient transmission in real-world applications. In this paper, we explore methods for performing semantic segmentation on the discrete cosine transform (DCT) representation defined by the JPEG standard. We first rearrange the DCT coefficients to form a preferred input type, then we tailor an existing network to the DCT inputs. The proposed method has an accuracy close to the RGB model at about the same network complexity. Moreover, we investigate the impact of selecting different DCT components on segmentation performance. With a proper selection, one can achieve the same level accuracy using only 36% of the DCT coefficients. We further show the robustness of our method under the quantization errors. To our knowledge, this paper is the first to explore semantic segmentation on the DCT representation.