Learning Classifiers for Domain Adaptation, Zero and Few-Shot Recognition Based on Learning Latent Semantic Parts

TL;DR

This paper introduces a novel visual attribute encoding method that improves domain adaptation, zero-shot, and few-shot recognition by representing images as low-dimensional probability vectors of prototypical parts, enhancing classifier adaptation.

Contribution

The paper proposes a new encoding technique based on learning representative part-type prototypes, enabling effective adaptation in low-data and zero-data scenarios.

Findings

Outperforms state-of-the-art methods in ZSL, FSL, and DA benchmarks.

Effective encoding of images as probability vectors of semantic parts.

Facilitates classifier adaptation with limited labeled data.

Abstract

In computer vision applications, such as domain adaptation (DA), few shot learning (FSL) and zero-shot learning (ZSL), we encounter new objects and environments, for which insufficient examples exist to allow for training "models from scratch," and methods that adapt existing models, trained on the presented training environment, to the new scenario are required. We propose a novel visual attribute encoding method that encodes each image as a low-dimensional probability vector composed of prototypical part-type probabilities. The prototypes are learnt to be representative of all training data. At test-time we utilize this encoding as an input to a classifier. At test-time we freeze the encoder and only learn/adapt the classifier component to limited annotated labels in FSL; new semantic attributes in ZSL. We conduct extensive experiments on benchmark datasets. Our method outperforms state-of-art methods trained for the specific contexts (ZSL, FSL, DA).