Rail Vehicle Interior Object Detection Using Masked Autoencoding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for detecting objects within vehicle interiors, particularly in rail vehicles, are inefficient and unreliable in recognizing objects that are not part of the permanently installed equipment, such as people, animals, or luggage, without requiring significant computational resources.

Innovation Solution

A computer-implemented method using a trained autoencoder with an encoder and decoder to analyze vehicle interior images, where image parts are masked and filled with filler data, followed by decoding and comparison to recognize objects, with optional averaging and feature clustering to enhance detection accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If trained neural networks are used to recognize objects in images, then object recognition capability is improved, but computational resource requirements increase

Engineering Contradiction:
Improveobject recognition reliabilityVSAvoidcomputational resource consumption
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The image is divided into multiple image parts, and only a subset of these parts is processed by the encoder at each step. This segmentation approach reduces the computational load on the neural network while maintaining effective object detection capabilities through multiple processing passes.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Instead of processing all image parts simultaneously with a full neural network, the method processes only a subset of image parts through the encoder in each iteration. Multiple iterations with different subsets achieve complete coverage with reduced computational resource consumption at each step.

Inventive Principle:
Principle #16Partial or excessive action

2Measurement precision

If multiple decoded images are generated and averaged, then recognition accuracy is improved, but processing time increases

Engineering Contradiction:
Improveobject detection accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The detection process is performed periodically multiple times with different subsets of image parts, generating multiple decoded images. These periodic detections are then averaged to improve accuracy, balancing the trade-off between detection precision and processing time through structured repetition.

Inventive Principle:
Principle #19Periodic action

Solution Approach 2:

Multiple decoded images are generated in advance through repeated processing with different image part subsets. The averaging of these pre-generated images improves detection accuracy before final object recognition, preparing enhanced data for the comparison step.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4614455A1Method for detecting a predetermined object in the interior of a vehicle
Publication Date: 2025.09.10 SIEMENS MOBILITY GMBH
  • EP4614455A1 patent drawingFigure 1~2
  • EP4614455A1 patent drawingFigure 3~4
  • EP4614455A1 patent drawingFigure 5A~6

AI summary

The invention relates to a computer-implemented method for recognizing a given object in the interior of a vehicle, in particular a rail vehicle, wherein an image of the interior is analyzed using a trained autoencoder, wherein the autoencoder has an encoder and a decoder, wherein the image of the interior is divided into image parts, wherein a first subset of the image parts of the image is processed by the encoder and a hidden representation of the image is determined, wherein the hidden representation has a hidden image part for each image part of the first subset, wherein filler image parts are subsequently inserted into the hidden image parts at the positions of the masked and non-analyzed image parts, wherein the decoder determines a decoded image based on the hidden image parts and the filler image parts,wherein the decoded image is compared with the image of the interior and, depending on a result of the comparison, the specified object is recognized or not, and wherein, after recognition of the specified object, a message is output and/or a message is stored.