End-to-End Learned Scalable Multilayer Feature Compression for Machine Vision Tasks

We propose an end-to-end learned scalable multilayer feature compression method. Our proposed method is illustrated in Figure 1 , where f 1 ,… ,f n stand for deep features at different layers. The deep feature f n denoting base layer is first transformed and quantized into the latent yₙ . The latent yₙ is then inversely transformed to reconstruct the feature as f ˆ n . In addition, the latent yₙ is also fed into the entropy model of the previous-layer feature f n−1 as conditional information for the enhancement layer. The entropy model of the feature f n−1 takes both yn - 1 and yₙ as inputs to improve compression efficiency.