Pixel Distillation: A New Knowledge Distillation Scheme for Low-Resolution Image Recognition

TL;DR

Pixel Distillation introduces a novel knowledge distillation approach that transfers spatial knowledge at the input level and employs a TAS framework, enabling flexible resource-aware deployment for image classification and object detection.

Contribution

It extends knowledge distillation to the input image level and proposes a TAS framework, allowing resource-efficient deployment and improved knowledge transfer between CNN and ViT.

Findings

Effective in image classification tasks

Improves object detection performance

Enables flexible resource control

Abstract

Previous knowledge distillation (KD) methods mostly focus on compressing network architectures, which is not thorough enough in deployment as some costs like transmission bandwidth and imaging equipment are related to the image size. Therefore, we propose Pixel Distillation that extends knowledge distillation into the input level while simultaneously breaking architecture constraints. Such a scheme can achieve flexible cost control for deployment, as it allows the system to adjust both network architecture and image quality according to the overall requirement of resources. Specifically, we first propose an input spatial representation distillation (ISRD) mechanism to transfer spatial knowledge from large images to student's input module, which can facilitate stable knowledge transfer between CNN and ViT. Then, a Teacher-Assistant-Student (TAS) framework is further established to disentangle pixel distillation into the model compression stage and input compression stage, which significantly reduces the overall complexity of pixel distillation and the difficulty of distilling intermediate knowledge. Finally, we adapt pixel distillation to object detection via an aligned feature for preservation (AFP) strategy for TAS, which aligns output dimensions of detectors at each stage by manipulating features and anchors of the assistant. Comprehensive experiments on image classification and object detection demonstrate the effectiveness of our method. Code is available at https://github.com/gyguo/PixelDistillation.