SMS3D: 3D Synthetic Mushroom Scenes Dataset for 3D Object Detection and Pose Estimation

The mushroom farming industry struggles to automate harvesting due to limited large-scale annotated datasets and the complex growth patterns of mushrooms, which complicate detection, segmentation, and pose estimation. To address this, we introduce a synthetic dataset with 40,000 unique scenes of whi...

Full description

Saved in:
Bibliographic Details
Main Authors: Abdollah Zakeri, Bikram Koirala, Jiming Kang, Venkatesh Balan, Weihang Zhu, Driss Benhaddou, Fatima A. Merchant
Format: Article
Language:English
Published: MDPI AG 2025-04-01
Series:Computers
Subjects:
Online Access:https://www.mdpi.com/2073-431X/14/4/128
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850144467922714624
author Abdollah Zakeri
Bikram Koirala
Jiming Kang
Venkatesh Balan
Weihang Zhu
Driss Benhaddou
Fatima A. Merchant
author_facet Abdollah Zakeri
Bikram Koirala
Jiming Kang
Venkatesh Balan
Weihang Zhu
Driss Benhaddou
Fatima A. Merchant
author_sort Abdollah Zakeri
collection DOAJ
description The mushroom farming industry struggles to automate harvesting due to limited large-scale annotated datasets and the complex growth patterns of mushrooms, which complicate detection, segmentation, and pose estimation. To address this, we introduce a synthetic dataset with 40,000 unique scenes of white <i>Agaricus bisporus</i> and brown baby bella mushrooms, capturing realistic variations in quantity, position, orientation, and growth stages. Our two-stage pose estimation pipeline combines 2D object detection and instance segmentation with a 3D point cloud-based pose estimation network using a Point Transformer. By employing a continuous 6D rotation representation and a geodesic loss, our method ensures precise rotation predictions. Experiments show that processing point clouds with 1024 points and the 6D Gram–Schmidt rotation representation yields optimal results, achieving an average rotational error of <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>1.67</mn><mo>°</mo></mrow></semantics></math></inline-formula> on synthetic data, surpassing current state-of-the-art methods in mushroom pose estimation. The model, further, generalizes well to real-world data, attaining a mean angle difference of <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>3.68</mn><mo>°</mo></mrow></semantics></math></inline-formula> on a subset of the M18K dataset with ground-truth annotations. This approach aims to drive automation in harvesting, growth monitoring, and quality assessment in the mushroom industry.
format Article
id doaj-art-3d032906aad047b4b71b3bb0ffb4f02d
institution OA Journals
issn 2073-431X
language English
publishDate 2025-04-01
publisher MDPI AG
record_format Article
series Computers
spelling doaj-art-3d032906aad047b4b71b3bb0ffb4f02d2025-08-20T02:28:20ZengMDPI AGComputers2073-431X2025-04-0114412810.3390/computers14040128SMS3D: 3D Synthetic Mushroom Scenes Dataset for 3D Object Detection and Pose EstimationAbdollah Zakeri0Bikram Koirala1Jiming Kang2Venkatesh Balan3Weihang Zhu4Driss Benhaddou5Fatima A. Merchant6Department of Computer Science, University of Houston, Houston, TX 77004, USADepartment of Engineering Technology, University of Houston, Houston, TX 77004, USADepartment of Engineering Technology, University of Houston, Houston, TX 77004, USADepartment of Engineering Technology, University of Houston, Houston, TX 77004, USADepartment of Engineering Technology, University of Houston, Houston, TX 77004, USADepartment of Engineering Technology, University of Houston, Houston, TX 77004, USADepartment of Computer Science, University of Houston, Houston, TX 77004, USAThe mushroom farming industry struggles to automate harvesting due to limited large-scale annotated datasets and the complex growth patterns of mushrooms, which complicate detection, segmentation, and pose estimation. To address this, we introduce a synthetic dataset with 40,000 unique scenes of white <i>Agaricus bisporus</i> and brown baby bella mushrooms, capturing realistic variations in quantity, position, orientation, and growth stages. Our two-stage pose estimation pipeline combines 2D object detection and instance segmentation with a 3D point cloud-based pose estimation network using a Point Transformer. By employing a continuous 6D rotation representation and a geodesic loss, our method ensures precise rotation predictions. Experiments show that processing point clouds with 1024 points and the 6D Gram–Schmidt rotation representation yields optimal results, achieving an average rotational error of <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>1.67</mn><mo>°</mo></mrow></semantics></math></inline-formula> on synthetic data, surpassing current state-of-the-art methods in mushroom pose estimation. The model, further, generalizes well to real-world data, attaining a mean angle difference of <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>3.68</mn><mo>°</mo></mrow></semantics></math></inline-formula> on a subset of the M18K dataset with ground-truth annotations. This approach aims to drive automation in harvesting, growth monitoring, and quality assessment in the mushroom industry.https://www.mdpi.com/2073-431X/14/4/128synthetic datasetmushroom detectionmushroom pose estimationpoint clouddeep learningcomputer vision
spellingShingle Abdollah Zakeri
Bikram Koirala
Jiming Kang
Venkatesh Balan
Weihang Zhu
Driss Benhaddou
Fatima A. Merchant
SMS3D: 3D Synthetic Mushroom Scenes Dataset for 3D Object Detection and Pose Estimation
Computers
synthetic dataset
mushroom detection
mushroom pose estimation
point cloud
deep learning
computer vision
title SMS3D: 3D Synthetic Mushroom Scenes Dataset for 3D Object Detection and Pose Estimation
title_full SMS3D: 3D Synthetic Mushroom Scenes Dataset for 3D Object Detection and Pose Estimation
title_fullStr SMS3D: 3D Synthetic Mushroom Scenes Dataset for 3D Object Detection and Pose Estimation
title_full_unstemmed SMS3D: 3D Synthetic Mushroom Scenes Dataset for 3D Object Detection and Pose Estimation
title_short SMS3D: 3D Synthetic Mushroom Scenes Dataset for 3D Object Detection and Pose Estimation
title_sort sms3d 3d synthetic mushroom scenes dataset for 3d object detection and pose estimation
topic synthetic dataset
mushroom detection
mushroom pose estimation
point cloud
deep learning
computer vision
url https://www.mdpi.com/2073-431X/14/4/128
work_keys_str_mv AT abdollahzakeri sms3d3dsyntheticmushroomscenesdatasetfor3dobjectdetectionandposeestimation
AT bikramkoirala sms3d3dsyntheticmushroomscenesdatasetfor3dobjectdetectionandposeestimation
AT jimingkang sms3d3dsyntheticmushroomscenesdatasetfor3dobjectdetectionandposeestimation
AT venkateshbalan sms3d3dsyntheticmushroomscenesdatasetfor3dobjectdetectionandposeestimation
AT weihangzhu sms3d3dsyntheticmushroomscenesdatasetfor3dobjectdetectionandposeestimation
AT drissbenhaddou sms3d3dsyntheticmushroomscenesdatasetfor3dobjectdetectionandposeestimation
AT fatimaamerchant sms3d3dsyntheticmushroomscenesdatasetfor3dobjectdetectionandposeestimation