Beyond Handcrafted Features: A Deep Learning Framework for Optical Flow and SLAM

This paper presents a novel approach for visual Simultaneous Localization and Mapping (SLAM) using Convolution Neural Networks (CNNs) for robust map creation. Traditional SLAM methods rely on handcrafted features, which are susceptible to viewpoint changes, occlusions, and illumination variations. T...

Full description

Saved in:
Bibliographic Details
Main Authors: Kamran Kazi, Arbab Nighat Kalhoro, Farida Memon, Azam Rafique Memon, Muddesar Iqbal
Format: Article
Language:English
Published: MDPI AG 2025-05-01
Series:Journal of Imaging
Subjects:
Online Access:https://www.mdpi.com/2313-433X/11/5/155
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849327012454334464
author Kamran Kazi
Arbab Nighat Kalhoro
Farida Memon
Azam Rafique Memon
Muddesar Iqbal
author_facet Kamran Kazi
Arbab Nighat Kalhoro
Farida Memon
Azam Rafique Memon
Muddesar Iqbal
author_sort Kamran Kazi
collection DOAJ
description This paper presents a novel approach for visual Simultaneous Localization and Mapping (SLAM) using Convolution Neural Networks (CNNs) for robust map creation. Traditional SLAM methods rely on handcrafted features, which are susceptible to viewpoint changes, occlusions, and illumination variations. This work proposes a method that leverages the power of CNNs by extracting features from an intermediate layer of a pre-trained model for optical flow estimation. We conduct an extensive search for optimal features by analyzing the offset error across thousands of combinations of layers and filters within the CNN. This analysis reveals a specific layer and filter combination that exhibits minimal offset error while still accounting for viewpoint changes, occlusions, and illumination variations. These features, learned by the CNN, are demonstrably robust to environmental challenges that often hinder traditional handcrafted features in SLAM tasks. The proposed method is evaluated on six publicly available datasets that are widely used for bench-marking map estimation and accuracy. Our method consistently achieved the lowest offset error compared to traditional handcrafted feature-based approaches on all six datasets. This demonstrates the effectiveness of CNN-derived features for building accurate and robust maps in diverse environments.
format Article
id doaj-art-9e8260787e8a4fdf9eaeeb76ca71b48a
institution Kabale University
issn 2313-433X
language English
publishDate 2025-05-01
publisher MDPI AG
record_format Article
series Journal of Imaging
spelling doaj-art-9e8260787e8a4fdf9eaeeb76ca71b48a2025-08-20T03:47:59ZengMDPI AGJournal of Imaging2313-433X2025-05-0111515510.3390/jimaging11050155Beyond Handcrafted Features: A Deep Learning Framework for Optical Flow and SLAMKamran Kazi0Arbab Nighat Kalhoro1Farida Memon2Azam Rafique Memon3Muddesar Iqbal4Institute of Information and Communication Technologies, Mehran University of Engineering and Technology, Jamshoro 76062, PakistanInstitute of Information and Communication Technologies, Mehran University of Engineering and Technology, Jamshoro 76062, PakistanDepartment of Electronic Engineering, Mehran University of Engineering and Technology, Jamshoro 76062, PakistanDepartment of Electronic Engineering, Mehran University of Engineering and Technology, Jamshoro 76062, PakistanRenewable Energy Laboratory, College of Engineering, Prince Sultan University, Riyadh 11586, Saudi ArabiaThis paper presents a novel approach for visual Simultaneous Localization and Mapping (SLAM) using Convolution Neural Networks (CNNs) for robust map creation. Traditional SLAM methods rely on handcrafted features, which are susceptible to viewpoint changes, occlusions, and illumination variations. This work proposes a method that leverages the power of CNNs by extracting features from an intermediate layer of a pre-trained model for optical flow estimation. We conduct an extensive search for optimal features by analyzing the offset error across thousands of combinations of layers and filters within the CNN. This analysis reveals a specific layer and filter combination that exhibits minimal offset error while still accounting for viewpoint changes, occlusions, and illumination variations. These features, learned by the CNN, are demonstrably robust to environmental challenges that often hinder traditional handcrafted features in SLAM tasks. The proposed method is evaluated on six publicly available datasets that are widely used for bench-marking map estimation and accuracy. Our method consistently achieved the lowest offset error compared to traditional handcrafted feature-based approaches on all six datasets. This demonstrates the effectiveness of CNN-derived features for building accurate and robust maps in diverse environments.https://www.mdpi.com/2313-433X/11/5/155CNN optical flow estimationfeature extractionmap buildingoffset errorvisual SLAM
spellingShingle Kamran Kazi
Arbab Nighat Kalhoro
Farida Memon
Azam Rafique Memon
Muddesar Iqbal
Beyond Handcrafted Features: A Deep Learning Framework for Optical Flow and SLAM
Journal of Imaging
CNN optical flow estimation
feature extraction
map building
offset error
visual SLAM
title Beyond Handcrafted Features: A Deep Learning Framework for Optical Flow and SLAM
title_full Beyond Handcrafted Features: A Deep Learning Framework for Optical Flow and SLAM
title_fullStr Beyond Handcrafted Features: A Deep Learning Framework for Optical Flow and SLAM
title_full_unstemmed Beyond Handcrafted Features: A Deep Learning Framework for Optical Flow and SLAM
title_short Beyond Handcrafted Features: A Deep Learning Framework for Optical Flow and SLAM
title_sort beyond handcrafted features a deep learning framework for optical flow and slam
topic CNN optical flow estimation
feature extraction
map building
offset error
visual SLAM
url https://www.mdpi.com/2313-433X/11/5/155
work_keys_str_mv AT kamrankazi beyondhandcraftedfeaturesadeeplearningframeworkforopticalflowandslam
AT arbabnighatkalhoro beyondhandcraftedfeaturesadeeplearningframeworkforopticalflowandslam
AT faridamemon beyondhandcraftedfeaturesadeeplearningframeworkforopticalflowandslam
AT azamrafiquememon beyondhandcraftedfeaturesadeeplearningframeworkforopticalflowandslam
AT muddesariqbal beyondhandcraftedfeaturesadeeplearningframeworkforopticalflowandslam