On-the-fly Network Pruning for Object Detection

TL;DR

This paper introduces an on-the-fly network pruning method for object detection that leverages feature occurrence at the image level to eliminate units with near-zero activation, significantly reducing network size with minimal impact on detection accuracy.

Contribution

It presents a novel image-scale feature-based pruning technique that dynamically reduces neural network complexity during object detection.

Findings

Up to 40% of units in fully-connected layers can be pruned.

Pruning causes minimal change in detection performance.

Method effectively reduces network size for faster inference.

Abstract

Object detection with deep neural networks is often performed by passing a few thousand candidate bounding boxes through a deep neural network for each image. These bounding boxes are highly correlated since they originate from the same image. In this paper we investigate how to exploit feature occurrence at the image scale to prune the neural network which is subsequently applied to all bounding boxes. We show that removing units which have near-zero activation in the image allows us to significantly reduce the number of parameters in the network. Results on the PASCAL 2007 Object Detection Challenge demonstrate that up to 40% of units in some fully-connected layers can be entirely eliminated with little change in the detection result.