3D Localization Research Articles

Optimal planning and management strategies for minimally invasive lung segmentectomies: an international Delphi consensus report.

CALGB140503/JCOG0802 RCTs comparing lobectomy with sublobar resection in stage IA NSCLC have confirmed the non-inferiority of segmentectomy. Additional insight is needed to improve preoperative work-up and intraoperative strategies to increase safety and promote the dissemination of minimally invasive segmentectomy (MIS). A Delphi panel study assessed the level of consensus among surgeons for the planning and management of MIS. Twenty-one expert lung surgeons represented academic institutions, major teaching hospitals, and community hospitals from Europe, North America, and Asia. A 3-round Delphi methodology was used to analyse the answers of each panellist. Recognizing that questions with fewer response options have a higher consensus probability due to limited variability, weighted consensus thresholds were modified based on the number of response options. The 21 panellists responded to all 3 rounds of questions. Based on the most robust consensus (94.4%), 3D chest CT reconstructions are recommended only when planning complex segmentectomies. Surgeons should perform 3D reconstructions chest CT scans (consensus = 83.3%). The most effective and safest technique is image-guided VATS in a hybrid operating room (consensus = 83.3%). Dyes with intravenous administration are the safest technique for identifying the intersegmental plane during MIS (consensus = 72.2%). Augmented/mixed reality will probably not immediately help reduce perioperative complications (consensus = 72.2%). This Delphi consensus supports 3D reconstructions and preoperative pulmonary nodule localization before complex MIS. These recommendations should be considered when allocating resources to improve MIS's safety and oncologic efficacy for patients with small, early-stage lung cancers.

Subject-Agnostic Transformer-Based Neural Speech Decoding from Surface and Depth Electrode Signals.

This study investigates speech decoding from neural signals captured by intracranial electrodes. Most prior works can only work with electrodes on a 2D grid (i.e., Electrocorticographic or ECoG array) and data from a single patient. We aim to design a deep-learning model architecture that can accommodate both surface (ECoG) and depth (stereotactic EEG or sEEG) electrodes. The architecture should allow training on data from multiple participants with large variability in electrode placements and the trained model should perform well on participants unseen during training. We propose a novel transformer-based model architecture named SwinTW that can work with arbitrarily positioned electrodes by leveraging their 3D locations on the cortex rather than their positions on a 2D grid. We train subject-specific models using data from a single participant and multi-patient models exploiting data from multiple participants. The subject-specific models using only low-density 8×8 ECoG data achieved high decoding Pearson Correlation Coefficient with ground truth spectrogram (PCC=0.817), over N=43 participants, outperforming our prior convolutional ResNet model and the 3D Swin transformer model. Incorporating additional strip, depth, and grid electrodes available in each participant (N=39) led to further improvement (PCC=0.838). For participants with only sEEG electrodes (N=9), subject-specific models still enjoy comparable performance with an average PCC=0.798. The multi-subject models achieved high performance on unseen participants, with an average PCC=0.765 in leave-one-out cross-validation. The proposed SwinTW decoder enables future speech neuropros-theses to utilize any electrode placement that is clinically optimal or feasible for a particular participant, including using only depth electrodes, which are more routinely implanted in chronic neurosurgical procedures. Importantly, the generalizability of the multi-patient models suggests that such a model can be applied to new patients that do not have paired acoustic and neural data, providing an advance in neuroprostheses for people with speech disability, where acoustic-neural training data is not feasible.

Observation of flat bands and Dirac cones in a pyrochlore lattice superconductor

Emergent phases often appear when the electronic kinetic energy is comparable to the Coulomb interactions. One approach to seek material systems as hosts of such emergent phases is to realize localization of electronic wavefunctions due to the geometric frustration inherent in the crystal structure, resulting in flat electronic bands. Recently, such efforts have found a wide range of exotic phases in the two-dimensional kagome lattice, including magnetic order, time-reversal symmetry breaking charge order, nematicity, and superconductivity. However, the interlayer coupling of the kagome layers disrupts the destructive interference needed to completely quench the kinetic energy. Here we demonstrate that an interwoven kagome network—a pyrochlore lattice—can host a three dimensional (3D) localization of electron wavefunctions. Meanwhile, the nonsymmorphic symmetry of the pyrochlore lattice guarantees all band crossings at the Brillouin zone X point to be 3D gapless Dirac points, which was predicted theoretically but never yet observed experimentally. Through a combination of angle-resolved photoemission spectroscopy, fundamental lattice model and density functional theory calculations, we investigate the novel electronic structure of a Laves phase superconductor with a pyrochlore sublattice, CeRu2. We observe evidence of flat bands originating from the Ce 4f orbitals as well as flat bands from the 3D destructive interference of the Ru 4d orbitals. We further observe the nonsymmorphic symmetry-protected 3D gapless Dirac cone at the X point. Our work establishes the pyrochlore structure as a promising lattice platform to realize and tune novel emergent phases intertwining topology and many-body interactions.

An innovative approach to 2D localization using the magnetic shielding effect

The effect on an electromagnetic field when a low-cost magnetically permeable object such as copper or aluminum is placed within it can be observed to determine the object’s location. This approach offers a novel technique to achieve reliable localization, particularly in environments where line of sight sensing methods may be non-applicable. Shields up to a size of 30×30 mm and a thickness of 80 µm were investigated; copper shields of these dimensions reduced the signal strength to 91 %, and aluminum shields reduced the signal strength to 94 % of its initial strength. The distortions to the electromagnetic field were closely related to the location of the tag. By fitting an inverted Gaussian curve to each sensor’s data, the position of a shield along a line could be predicted. This method can be used to locate a tag within a 2D plane by creating a 2D array of sensors beneath the sensing plane.

LILOC: Leveraging LiDARs for Accurate 3D Localization in Dynamic Indoor Environments

LILOC: Leveraging LiDARs for Accurate 3D Localization in Dynamic Indoor Environments

ECLiPSE: a versatile classification technique for structural and morphological analysis of 2D and 3D single-molecule localization microscopy data

Single-molecule localization microscopy (SMLM) has gained widespread use for visualizing the morphology of subcellular organelles and structures with nanoscale spatial resolution. However, analysis tools for automatically quantifying and classifying SMLM images have lagged behind. Here we introduce Enhanced Classification of Localized Point clouds by Shape Extraction (ECLiPSE), an automated machine learning analysis pipeline specifically designed to classify cellular structures captured through two-dimensional or three-dimensional SMLM. ECLiPSE leverages a comprehensive set of shape descriptors, the majority of which are directly extracted from the localizations to minimize bias during the characterization of individual structures. ECLiPSE has been validated using both unsupervised and supervised classification on datasets, including various cellular structures, achieving near-perfect accuracy. We apply two-dimensional ECLiPSE to classify morphologically distinct protein aggregates relevant for neurodegenerative diseases. Additionally, we employ three-dimensional ECLiPSE to identify relevant biological differences between healthy and depolarized mitochondria. ECLiPSE will enhance the way we study cellular structures across various biological contexts.

AI-Enabled Sensor Fusion of Time-of-Flight Imaging and mmWave for Concealed Metal Detection.

In the field of detection and ranging, multiple complementary sensing modalities may be used to enrich information obtained from a dynamic scene. One application of this sensor fusion is in public security and surveillance, where efficacy and privacy protection measures must be continually evaluated. We present a novel deployment of sensor fusion for the discrete detection of concealed metal objects on persons whilst preserving their privacy. This is achieved by coupling off-the-shelf mmWave radar and depth camera technology with a novel neural network architecture that processes radar signals using convolutional Long Short-Term Memory (LSTM) blocks and depth signals using convolutional operations. The combined latent features are then magnified using deep feature magnification to reveal cross-modality dependencies in the data. We further propose a decoder, based on the feature extraction and embedding block, to learn an efficient upsampling of the latent space to locate the concealed object in the spatial domain through radar feature guidance. We demonstrate the ability to detect the presence and infer the 3D location of concealed metal objects. We achieve accuracies of up to 95% using a technique that is robust to multiple persons. This work provides a demonstration of the potential for cost-effective and portable sensor fusion with strong opportunities for further development.

Localization and recognition of human action in 3D using transformers

Understanding a person’s behavior from their 3D motion sequence is a fundamental problem in computer vision with many applications. An important component of this problem is 3D action localization, which involves recognizing what actions a person is performing, and when the actions occur in the sequence. To promote the progress of the 3D action localization community, we introduce a new, challenging, and more complex benchmark dataset, BABEL-TAL (BT), for 3D action localization. Important baselines and evaluating metrics, as well as human evaluations, are carefully established on this benchmark. We also propose a strong baseline model, i.e., Localizing Actions with Transformers (LocATe), that jointly localizes and recognizes actions in a 3D sequence. The proposed LocATe shows superior performance on BABEL-TAL as well as on the large-scale PKU-MMD dataset, achieving state-of-the-art performance by using only 10% of the labeled training data. Our research could advance the development of more accurate and efficient systems for human behavior analysis, with potential applications in areas such as human-computer interaction and healthcare.

Knowing where to go: Spatial memory guides eye and body movements in a naturalistic visual search task.

Most research on visual search has used simple tasks presented on a computer screen. However, in natural situations visual search almost always involves eye, head, and body movements in a three-dimensional (3D) environment. The different constraints imposed by these two types of search tasks might explain some of the discrepancies in our understanding concerning the use of memory resources and the role of contextual objects during search. To explore this issue, we analyzed a visual search task performed in an immersive virtual reality apartment. Participants searched for a series of geometric 3D objects while eye movements and head coordinates were recorded. Participants explored the apartment to locate target objects whose location and visibility were manipulated. For objects with reliable locations, we found that repeated searches led to a decrease in search time and number of fixations and to a reduction of errors. Searching for those objects that had been visible in previous trials but were only tested at the end of the experiment was also easier than finding objects for the first time, indicating incidental learning of context. More importantly, we found that body movements showed changes that reflected memory for target location: trajectories were shorter and movement velocities were higher, but only for those objects that had been searched for multiple times. We conclude that memory of 3D space and target location is a critical component of visual search and also modifies movement kinematics. In natural search, memory is used to optimize movement control and reduce energetic costs.

Three-dimensional grid-free sound source localization method based on deep learning

Sound source localization (SSL) technology is a popular method for identifying the locations of noise sources, which serves as a prerequisite for noise control. Deep learning, as a data-driven tool, shows broad perspectives in the field of SSL with its powerful nonlinear fitting ability. The existing deep learning-based SSL methods only provide a two-dimensional (2D) representation of the sound source location and cannot obtain the specific coordinates of the sound source in three-dimensional (3D) space. Although traditional beamforming methods can be directly generalized to 3D scenes in principle, they suffer from the limitations of insufficient vertical resolution and high computational cost. Therefore, a 3D grid-free SSL method (3DGF) informed by deep learning is suggested in this study to enhance the accuracy and computational efficiency of 3D localization. First, the number of data channels is compressed to respect limited memory resources during the training process. Subsequently, a dense convolutional neural network (DenseNet) model is utilized to obtain the 3D spatial coordinates of the sound source using the processed 3D beamforming map as input. Since the coordinates are continuous and are not constrained by the grid of the beamforming map, the grid-free strategy presents more accurate localization results. Then, the effects of the volume of training data and the compression ratio are analyzed, respectively, in simulation, and the localization performance with different signal-to-noise ratios (SNRs) is also tested. Finally, by comparing 3DGF with DAMAS, both simulation and experimental results demonstrate that 3DGF improves the accuracy and efficacy of 3D localization. Meanwhile, its satisfactory generalization ability and robustness against noise highlight its potential for practical applications.

Rapid flowing cells localization enabled by spatiotemporal manipulation of their holographic patterns.

Lab-on-a-Chip microfluidic devices present an innovative and cost-effective platform in the current trend of miniaturization and simplification of imaging flow cytometry; they are excellent candidates for high-throughput single-cell analysis. In such microfluidic platforms, cell tracking becomes a fundamental tool for investigating biophysical processes, from intracellular dynamics to the characterization of cell motility and migration. However, high-throughput and long-term cell tracking puts a high demand on the consumption of computing resources. Here, we propose a novel strategy to achieve rapid 3D cell localizations along the microfluidic channel. This method is based on the spatiotemporal manipulation of recorded holographic interference fringes, and it allows fast and precise localization of cells without performing complete holographic reconstruction. Conventional holographic tracking is typically based on the phase contrast obtained by decoupling the calculation of optical axial and transverse coordinates. Computing time and resource consumption may increase because all the frames need to be calculated in the Fourier domain. In our proposed method, the 2D transverse positions are directly located by morphological calculation based on the hologram. The complex-amplitude wavefronts are directly reconstructed by spatiotemporal phase shifting to calculate the axial position by the refocusing criterion. Only spatial calculation is considered in the proposed method. We demonstrate that the computational time of transverse tracking is only one-tenth of the conventional method, while the total computational time of the proposed method decreases up to 54% with respect to the conventional approach. The proposed approach can open the route for analyzing flow cytometry in quantitative phase microscopy assays.

An efficient and lightweight banana detection and localization system based on deep CNNs for agricultural robots

Accurate detection and localization of fruits in natural environments is a key step for fruit picking robots to achieve precise harvesting. However, existing banana detection and positioning methods have two main limitations in practical applications: a large number of model parameters that make deployment difficult, and a need for performance improvement. To tackle the above issues, a high-precision and lightweight banana bunch recognition and localization method was proposed and deployed on edge devices for application. First, a Slim-Banana model was proposed based on the improvement of YOLOv8l. In order to reduce the model calculation amount and maintain high performance, GSConv was introduced in the Slim-Banana model to replace the standard convolution, and combined with grouped convolution and spatial convolution. At the same time, the cross-stage local network (GSCSP) module was designed to reduce the computational complexity and the complexity of the network structure through a single-stage aggregation method. Then, the RealSense depth sensor is combined with TOF technology to perform image registration and 3D localization of the banana. Finally, the pipeline is deployed on the Nvidia Orin NX edge device and its performance and resource consumption in actual work are deeply analyzed. Experimental results show that the detection precision, recall, mAP and inference time of our method are 0.947, 0.948, 0.98 and 113.6ms respectively, the network memory size required is 4449MiB, and the average localization errors in the X-axis, Y-axis and Z-axis directions are 13.47mm, 12.87mm and 13.87mm respectively. To our knowledge, this is the first work that implements banana detection and localization on edge devices. Experimental results show that compared with existing methods, our method achieves better performance in complex orchard environments, achieving efficient and lightweight banana recognition and localization.

Optimized intelligent 3D localization in wireless sensor networks for better data sharing

WSNs are networks of small, autonomous devices called sensors or nodes that are equipped with sensors, processing capabilities, and wireless communication capabilities. These networks are designed to monitor and collect data from their surrounding environments and transmit that data wirelessly to a central location for analysis. However, challenges such as packet loss and limited throughput have impeded their efficiency. To address these issues, this research presents an innovative approach known as Squirrel-based Elman Neural Localization (SbENL) to optimize 3D localization in WSNs. Squirrel fitness tracks the node's location in the sensor network environment. This study initially configures WSN nodes with constrained energy resources, and data rates are continually monitored and predicted. The outcome is improved data sharing with higher throughput rates. The research assesses the performance of SbENL against existing localization techniques, demonstrating its effectiveness in minimizing data loss, reducing energy consumption, and maximizing data transfer rates. This optimized intelligent 3D localization approach holds promise for enhancing data sharing and communication efficiency in WSNs, benefiting a wide range of applications. Finally, the communication metrics were measured and valued with recent existing approaches. A SbENL has minimized the data loss energy consumption and maximized the data transferring rate.

ReLoki: A Light-Weight Relative Localization System Based on UWB Antenna Arrays.

Ultra Wide-Band (UWB) sensing has gained popularity in relative localization applications. Many localization solutions rely on using Time of Flight (ToF) sensing based on a beacon-tag system, which requires four or more beacons in the environment for 3D localization. A lesser researched option is using Angle of Arrival (AoA) readings obtained from UWB antenna pairs to perform relative localization. In this paper, we present a UWB platform called ReLoki that can be used for ranging and AoA-based relative localization in 3D. To enable AoA, ReLoki utilizes the geometry of antenna arrays. In this paper, we present a system design for localization estimates using a Regular Tetrahedral Array (RTA), Regular Orthogonal Array (ROA), and Uniform Square Array (USA). The use of a multi-antenna array enables fully onboard infrastructure-free relative localization between participating ReLoki modules. We also present studies demonstrating sub-50cm localization errors in indoor experiments, achieving performance close to current ToF-based systems, while offering the advantage of not relying on static infrastructure.

Toward structure and metabolism of glycogen C1-C6 in humans at 7T by localized 13C MRS using low-power bilevel broadband 1H decoupling.

This work aims to develop and implement a pulse-acquire sequence for three-dimensional (3D) single-voxel localized 13C MRS in humans at 7T, in conjunction with bilevel broadband 1H decoupling, and to test its feasibility in vitro and in vivo in human calf muscle with emphasis on the detection of glycogen C1-C6. A localization scheme suitable for measuring fast-relaxing 13C signals in humans at 7T was developed and implemented using the outer volume suppression (OVS) and one-dimensional image selected in vivo spectroscopy (ISIS-1D) schemes, similar to that which was previously reported in humans at 4T. The 3D 13C localization scheme was followed by uniform 13C adiabatic excitation, all complemented with an option for bilevel broadband 1H decoupling to improve both 13C sensitivity and spectral resolution at 7T. The performance of the pulse-acquire sequence was investigated in vitro on phantoms and in vivo in the human calf muscle of three healthy volunteers, while measuring glycogen C1-C6. In addition, T1 and T2 of glycogen C1-C6 were measured in vitro at 7T, as well as T1 of glycogen C1 in vivo. The glycerol C2 and C1,3 lipid resonances were efficiently suppressed in vitro at 7T using the OVS and ISIS-1D schemes, allowing distinct detection of glycogen C2-C6. While some glycerol remained in calf muscle in vivo, the intense lipid at 130 ppm was efficiently suppressed. The 13C sensitivity and spectral resolution of glycogen C1-C6 in vitro and glycogen C1 in vivo were improved at 7T using bilevel broadband 1H decoupling. The T1 and T2 of glycogen C1-C6 in vitro at 7T were consistent compared with those at 8.5T, while the T1 of glycogen C1 in vivo at 7T resulted similar to that in vitro. Localized 13C MRS is feasible in human calf muscle in vivo at 7T, and this will allow further extension of this method for 13C MRS measurements such as in the brain.

Optimization of indoor vehicle ultra-wideband 3D localization using graph attention networks

Ultra-wideband technology is recognized as a low-cost and interference-resistant method for indoor vehicle localization, crucial for advancing intelligent transportation systems. Within this domain, the time difference of arrival algorithm finds extensive application. However, the accuracy of indoor three-dimensional (3D) localization is significantly affected by multipath propagation and non-line-of-sight deviations. To improve localization accuracy, this study proposes an innovative method utilizing the graph attention network framework. This method extracts relevant features and employs attention mechanisms to differentially weight individual nodes; it adeptly captures crucial spatial relationships among nodes, thereby mitigating the effects of multipath propagation and non-line-of-sight deviations, ultimately achieving finer indoor vehicle localization. Experimental results from tests on line-of-sight and non-line-of-sight data sets illustrate the significant improvement in indoor vehicle 3D localization accuracy afforded by the proposed graph attention network–based method. Compared to the commonly used built-in error-state Kalman filter algorithm in commercial ultra-wideband applications and alternative deep learning algorithms, the proposed method achieves error reductions ranging from 60.58% to 84.16% in line-of-sight environments and from 65.59% to 95.21% in non-line-of-sight environments.

Predictive power of deleterious single amino acid changes to infer on AAV2 and AAV2-13 capsids fitness

Adeno-associated virus (AAV) is the most widely used vector for invivo gene transfer. A major limitation of capsid engineering is the incomplete understanding of the consequences of multiple amino acid variations on AAV capsid stability resulting in high frequency of non-viable capsids. In this context, the study of natural AAV variants can provide valuable insights into capsid regions that exhibit greater tolerance to mutations. Here, the characterization of AAV2 variants and the analysis of two public capsid libraries highlighted common features associated with deleterious mutations, suggesting that the impact of mutations on capsid viability is strictly dependent on their 3D location within the capsid structure. We developed a novel prediction method to infer the fitness of AAV2 variants containing multiple amino acid variations with 98% sensitivity, 98% accuracy, and 95% specificity. This novel approach might streamline the development of AAV vector libraries enriched in viable capsids, thus accelerating the identification of therapeutic candidates among engineered capsids.

A Feasibility Study of HoloLens Ear Image Guidance for Ear Reconstruction.

An ideal 3D cartilage framework and accurate anatomical location are the most important factors to carry out a satisfactory reconstruction of the ear. To streamline this process, we developed an augmented reality assistance system, HoloLens Ear Image Guidance, which is based on computed tomography (CT) data, tailor-made navigation kits to guide reconstruction, data processing software, and HoloLens hardware. The objective of this study is to verify its feasibility in a clinical setting. This study first validated our system in healthy controls and then extrapolated data to test on patients with microtia. First, three healthy volunteers were recruited, and reconstructive navigation kits were made using 3D printing. CT data were collected for the head and neck and imported into the HoloLens Ear Image Guidance Application to generate a personalized 3D virtual ear image. Volunteers then wore the navigation kits while researchers observed them through the HoloLens to check accuracy, track delay, and view the ear image guide.Ten patients with unilateral microtia were recruited and CT data were collected, and reconstructive navigation kits were made to assist with surgery. The procedure was monitored to record the surgeon's experience wearing the HoloLens, the patients' complications associated with wearing navigation kit, and to measure the symmetry between the reconstructed ear and the reference counterpart. In control patients, the deviation between the virtual image and the real ear was less than 2.4% (±0.22%); the tracking delay was less than 1.26 s (±0.09 s), the display effect was good, and surgeons did not report discomfort or dizziness while wearing the HoloLens. Volunteers did not report any pain from holding the navigation reference in their mouth during the test. Following validation, the HoloLens-assisted procedures were not associated with surgeon discomfort or dizziness. No complications were noted in patients including injury to the oral mucosa. Symmetry between the reconstructed ear and the contralateral ear was noted to be satisfactory in HoloLens-assisted surgery. The HoloLens Ear Image Guidance initially met clinical demands in registration accuracy, tracking speed, and subjective user experience, which can be used as the basis for continual software improvements and clinical application.

The Emergence of the Slc11 Clade MCbgut: A Parsimonious Hypothesis for the Dawn of Lactobacillales in the Gut of Early Vertebrates

The Lactobacillales (LB) stand apart among bacterial orders, using manganese (Mn) instead of iron to support their growth and swiftly ferment complex foods while acidifying their environment. The present work investigates whether a shift in the use of Mn could mark the origin of LB. Transmembrane carriers of the ubiquitous Slc11 family play key roles in LB physiology by catalyzing proton-dependent Mn import. In prior studies, the Slc11 clade found in LB (MntH Cb, MCb) showed both remarkable structural plasticity and highly efficient Mn uptake, and another Slc11 clade, MCg1, demonstrated divergent evolution coinciding with emergence of bacterial genera (e.g., Bordetella, Achromobacter). Herein, the Slc11 clade MCb is subdivided in sister groups: MCbie and MCbgut. MCbie derives directly from the Slc11 clade MCa, pointing an intermediate stage in the evolution of MCbgut. MCbie predominates in marine Bacillaceae, is more conserved than MCbgut, lacks the structural plasticity that typify MCbgut carriers, and responds differently to identical mutagenesis. Exchanging MCbie/MCbgut amino acid residues at sites that distinguish these clades showed conformation-dependent effects with both MCbie and MCbgut templates, and the 3D location of the targeted sites in the carrier structure together suggests that the mechanism to open the inner gate, and release Mn into the cytoplasm, differs between MCbie and MCbgut. Building on the established phylogeny for Enterococcus revealed that a pair of genes encoding MCbgut was present in the common ancestor of LB, as MCbgu1 and MCbgu2 templates exhibited distinct structural dynamics properties. These data are discussed when examining whether MCbgut+ LB could emerge in the upper gut of early vertebrates (ca. 540 mya), through genome contraction and evolution toward Mn-centrism, as they specialized as gastric aids favoring stomach establishment in jawed vertebrates through bi-directional communication with host nervous, endocrine and immune systems.

S2CPL: A novel method of the harvest evaluation and subsoil 3D cutting-Point location for selective harvesting of green asparagus

Robotic selective harvesting is an ideal method for bionic manual harvest of green asparagus. However, the harvesting robot encounters difficulties in evaluating the suitable harvest due to the tilt and bending of the long stem, as well as determining the precise location of the subsoil cutting-point to prevent damage from bacteria on the cutting surface. This paper proposed the S2CPL model to address the challenges of the harvest evaluation and 3D localization of subsoil cutting-point for selective harvesting of green asparagus in field conditions. Firstly, an RGB-D sensor was used to acquire images and depth information of green asparaguses. Secondly, the improved YOLOv8 by introduced lightweight convolution and attention mechanisms in the feature fusion module to enhance the segmentation accuracy. Thirdly, a 3D morphology extraction method was proposed to calculate the length and diameter of green asparagus by utilizing the image mask fusion with depth information. Finally, harvest evaluation and subsoil 3D cutting-point location were achieved for robotic selective harvesting. In addition, the RGB-D sensor posture was optimized. The test results showed that the Intersection over Union (IoU) of green asparagus segmentation with S2CPL reaches 98.0 %, which outperforms YOLOv5 + uNet, YOLOv7 + uNet and YOLOv8-tiny by 5.60 %, 4.59 % and 1.34 % respectively. The average detection time per image was only 2.0 ms, and the GFLOPS was improved by 23.90 %, 88.49 % and 7.63 % compared with other models. The relative error of the length and diameter were less than 2.98 % and 2.15 %, respectively. The accuracy of location the subsoil cutting-point is more than 99.0 %, and the horizontal positioning error and depth positioning error of cutting-points were less than 6.0 mm and 7.4 mm. The proposed model is of strong robustness even dealing with partial occlusion and motion blur and is suitable with limited computing power to meet the needs of Robotic selective harvesting.