Distilling Image Classifiers in Object Detectors

Abstract

Knowledge distillation constitutes a simple yet effective way to improve the performance of a compact student network by exploiting the knowledge of a more powerful teacher. Nevertheless, the knowledge distillation literature remains limited to the scenario where the student and the teacher tackle the same task. Here, we investigate the problem of transferring knowledge not only across architectures but also across tasks. To this end, we study the case of object detection and, instead of following the standard detector-to-detector distillation approach, introduce a classifier-to-detector knowledge transfer framework. In particular, we propose strategies to exploit the classification teacher to improve both the detector's recognition accuracy and localization performance. Our experiments on several detectors with different backbones demonstrate the effectiveness of our approach, allowing us to outperform the state-of-the-art detector-to-detector distillation methods.

References in corpus (9)

• Distilling the Knowledge in a Neural Network
• MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
• YOLOv3: An Incremental Improvement
• Paying More Attention to Attention: Improving the Performance of Convolutional Neural Networks via Attention Transfer
• MMDetection: Open MMLab Detection Toolbox and Benchmark
• SNIP: Single-shot Network Pruning based on Connection Sensitivity
• Improving Neural Network Quantization without Retraining using Outlier Channel Splitting
• Compression-aware Training of Deep Networks
• General Instance Distillation for Object Detection