Gram-SLD: Automatic Self-labeling and Detection for Instance Objects

TL;DR

This paper introduces Gram-SLD, a co-training framework that automatically labels data for instance object detection, significantly reducing manual annotation effort while maintaining high detection accuracy in complex environments.

Contribution

The paper proposes a novel Gram-SLD framework that uses gram loss, view construction, and sample selection to generate high-quality pseudo-labels with minimal manual annotation.

Findings

Achieves less than 2% mAP loss with only 5% labeled data.

Demonstrates competitive performance on multiple datasets.

Satisfies real-time and accuracy requirements in complex environments.

Abstract

Instance object detection plays an important role in intelligent monitoring, visual navigation, human-computer interaction, intelligent services and other fields. Inspired by the great success of Deep Convolutional Neural Network (DCNN), DCNN-based instance object detection has become a promising research topic. To address the problem that DCNN always requires a large-scale annotated dataset to supervise its training while manual annotation is exhausting and time-consuming, we propose a new framework based on co-training called Gram Self-Labeling and Detection (Gram-SLD). The proposed Gram-SLD can automatically annotate a large amount of data with very limited manually labeled key data and achieve competitive performance. In our framework, gram loss is defined and used to construct two fully redundant and independent views and a key sample selection strategy along with an automatic annotating strategy that comprehensively consider precision and recall are proposed to generate high quality pseudo-labels. Experiments on the public GMU Kitchen Dataset , Active Vision Dataset and the self-made BHID-ITEM Datasetdemonstrate that, with only 5% labeled training data, our Gram-SLD achieves competitive performance in object detection (less than 2% mAP loss), compared with the fully supervised methods. In practical applications with complex and changing environments, the proposed method can satisfy the real-time and accuracy requirements on instance object detection.