Self-supervised Representation Learning for Reliable Robotic Monitoring of Fruit Anomalies (2021-09-21T00:00:00.000000Z)

TL;DR

Channel Randomisation is proposed as a novel data augmentation method for restricting neural networks to learn encoding of “colour irregularity” whilst predicting channel-randomised images to ultimately build reliable fruit-monitoring robots identifying atypical fruit qualities.

Abstract

Data augmentation can be a simple yet powerful tool for autonomous robots to fully utilise available data for self-supervised identification of atypical scenes or objects. State-of-the-art augmentation methods arbitrarily embed “structural” peculiarity on typical images so that classifying these artefacts can provide guidance for learning representations for the detection of anomalous visual signals. In this paper, however, we argue that learning such structure-sensitive representations can be a suboptimal approach to some classes of anomaly (e.g., unhealthy fruits) which could be better recognised by a different type of visual element such as “colour”. We thus propose Channel Randomisation as a novel data augmentation method for restricting neural networks to learn encoding of “colour irregularity” whilst predicting channel-randomised images to ultimately build reliable fruit-monitoring robots identifying atypical fruit qualities. Our experiments show that (1) this colour-based alternative can better learn representations for consistently accurate identification of fruit anomalies in various fruit species, and also, (2) unlike other methods, the validation accuracy can be utilised as a criterion for early stopping of training in practice due to positive correlation between the performance in the self-supervised colour-differentiation task and the subsequent detection rate of actual anomalous fruits. Also, the proposed approach is evaluated on a new agricultural dataset, Riseholme-2021, consisting of 3.5K strawberry images gathered by a mobile robot, which we share online to encourage active agri-robotics research.

Authors

References34 items

CutPaste: Self-Supervised Learning for Anomaly Detection and Localization

Combining generative adversarial networks and agricultural transfer learning for weeds identification

The MVTec Anomaly Detection Dataset: A Comprehensive Real-World Dataset for Unsupervised Anomaly Detection

Interactive Movement Primitives: Planning to Push Occluding Pieces for Fruit Picking

Identification of Abnormal States in Videos of Ants Undergoing Social Phase Change

Automatic late blight lesion recognition and severity quantification based on field imagery of diverse potato genotypes by deep learning

Classification of pickable and unpickable strawberries under farm conditions

Patch SVDD: Patch-level SVDD for Anomaly Detection and Segmentation

Modeling the Distribution of Normal Data in Pre-Trained Deep Features for Anomaly Detection

Robotic weed control using automated weed and crop classification

Self-supervised Label Augmentation via Input Transformations

Novelty Detection for Multispectral Images with Application to Planetary Exploration

Using Self-Supervised Learning Can Improve Model Robustness and Uncertainty

f‐AnoGAN: Fast unsupervised anomaly detection with generative adversarial networks

Transfer learning between crop types for semantic segmentation of crops versus weeds in precision agriculture

Multilayer Convolution Neural Network for the Classification of Mango Leaves Infected by Anthracnose Disease

OCGAN: One-Class Novelty Detection Using GANs With Constrained Latent Representations

A Color and Texture Based Approach for the Detection and Classification of Plant Leaf Disease Using KNN Classifier

Revisiting Self-Supervised Visual Representation Learning

Deep One-Class Classification

Tracking Emerges by Colorizing Videos

Adversarially Learned One-Class Classifier for Novelty Detection

Unsupervised Representation Learning by Predicting Image Rotations

Deep learning models for plant disease detection and diagnosis

Learning Deep Features for One-Class Classification

Improved Regularization of Convolutional Neural Networks with Cutout

Unsupervised Representation Learning by Sorting Sequences

Colorization as a Proxy Task for Visual Understanding

Very Deep Convolutional Networks for Large-Scale Image Recognition

ImageNet: A large-scale hierarchical image database

Image Segmentation

Estimating the Support of a High-Dimensional Distribution

Multilayer

Learning Multiple Layers of Features from Tiny Images