Approach For Object Recognition Research Articles

Synthetic lidar point cloud generation using deep generative models for improved driving scene object recognition

The imbalanced distribution of different object categories poses a challenge for training accurate object recognition models in driving scenes. Supervised machine learning models trained on imbalanced data are biased and easily overfit the majority classes, such as vehicles and pedestrians, which appear more frequently in driving scenes. We propose a novel data augmentation approach for object recognition in lidar point cloud of driving scenes, which leverages probabilistic generative models to produce synthetic point clouds for the minority classes and complement the original imbalanced dataset. We evaluate five generative models based on different statistical principles, including Gaussian mixture model, variational autoencoder, generative adversarial network, adversarial autoencoder and the diffusion model. Experiments with a real-world autonomous driving dataset show that the synthetic point clouds generated for the minority classes by the Latent Generative Adversarial Network result in significant improvement of object recognition performance for both minority and majority classes. The codes are available at https://github.com/AAAALEX-XIANG/Synthetic-Lidar-Generation.

Z-YOLOv8s-based approach for road object recognition in complex traffic scenarios

Object detection in road scenarios is crucial for intelligent transport systems and autonomous driving, but complex traffic conditions pose significant challenges. This paper introduces Z-You Only Look Once version 8 small (Z-YOLOv8s), designed to improve both accuracy and real-time efficiency under real-world uncertainties. By incorporating Revisiting Perspective Vision Transformer (RepViT) and C2f into the YOLOv8s framework, and integrating the Large Selective Kernel Network (LSKNet), the model enhances spatial feature extraction. Additionally, the YOLOv8s backbone is optimized with Space-to-Depth Convolution (SPD-Conv) for better small object detection. The Softpool-Spatial Pyramid Pooling Fast (SoftPool-SPPF) module ensures precise characteristic information preservation. Z-YOLOv8s improves mean average precision (mAP)@0.5 on the Berkeley Deep Drive 100 K (BDD100K) and Karlsruhe Institute of Technology and Toyota Technological Institute (KITTI) datasets by 7.3 % and 3.8 %, respectively. It also achieves accuracy increases of 5.7 % and 6.5 % in Average Precision (AP)-Small, and a real-time detection speed of 78.41 frames per second (FPS) on the BDD100K. Z-YOLOv8s balances detection precision and processing speed more effectively than other detectors, as demonstrated by experimental results and comparisons.

Leveraging computer vision towards high-efficiency autonomous industrial facilities

AbstractManufacturers face two opposing challenges: the escalating demand for customized products and the pressure to reduce delivery lead times. To address these expectations, manufacturers must refine their processes, to achieve highly efficient and autonomous operations. Current manufacturing equipment deployed in several facilities, while reliable and produces quality products, often lacks the ability to utilize advancements from newer technologies. Since replacing legacy equipment may be financially infeasible for many manufacturers, implementing digital transformation practices and technologies can overcome the stated deficiencies and offer cost-affordable initiatives to improve operations, increase productivity, and reduce costs. This paper explores the implementation of computer vision, as a cutting-edge, cost-effective, open-source digital transformation technology in manufacturing facilities. As a rapidly advancing technology, computer vision has the potential to transform manufacturing operations in general, and quality control in particular. The study integrates a digital twin application at the endpoint of an assembly line, effectively performing the role of a quality officer by utilizing state-of-the-art computer vision algorithms to validate end-product assembly orientation. The proposed digital twin, featuring a novel object recognition approach, efficiently classifies objects, identifies and segments errors in assembly, and schedules the paths through the data pipeline to the corresponding robot for autonomous correction. This minimizes the need for human interaction and reduces disruptions to manufacturing operations.

Odors Detection and Recognition Based on Intelligent E-Nose

Electronic noses have become more common as a result of developments in sensor technology, machine learning (ML), and Artificial Intelligence (AI). At the moment, the majority of e-nose research is conducted in laboratories; access from other locations is not possible with e-nose. Very few real-time smell detection applications, including tiny drones equipped with commercial gas sensors or biosensors made of insect antennas, have been created. The scope of this work is to design an intelligent E-Nose for odors detection. Smell Inspector developer kit has been used to get measurements. The Smell Inspector consists of sensors for temperature and humidity in addition to four smell iX16 chips. The Smell Inspector creates digital fragrance fingerprints using a variety of separate gas detectors. The approach for object recognition and classification using artificial intelligence techniques is presented in this paper. These techniques include Machine Learning (ML), clustering, and regression algorithms. Using the K-Nearest Neighbor (KNN) algorithm, the experimental results' array sensors were able to recognize; clean air, onion, garlic, coffee, spices, lemon, vinegar, gasoline, petrol, diesel, and perfumes with 78% accuracy rate.

Color-Driven Object Recognition: A Novel Approach Combining Color Detection and Machine Learning Techniques

INTRODUCTION: Object recognition is a crucial task in computer vision, with applications in robotics, autonomous vehicles, and security systems. 
 OBJECTIVES: The objective of this paper is to propose a novel approach for object recognition by combining color detection and machine learning techniques.
 METHODS: The research employs YOLO v3, a state-of-the-art object detection algorithm, and k-means optimized clustering to enhance the accuracy and efficiency of object recognition. RESULTS: The main results obtained in this paper showcase the outperformance of the authors’ approach on a standard object recognition dataset compared to state-of-the-art approaches using only color features. Additionally, the effectiveness of this approach is demonstrated in a real-world scenario of detecting and tracking objects in a video stream.
 CONCLUSION: In conclusion, this approach, integrating color and shape features, has the potential to significantly enhance the accuracy and robustness of object recognition systems. This contribution can pave the way for the development of more reliable and efficient object recognition systems across various applications.

Proposed Approach for Object Detection and Recognition by Deep Learning Models Using Data Augmentation

Object detection and recognition play a crucial role in computer vision applications, ranging from security systems to autonomous vehicles. Deep learning algorithms have shown remarkable performance in these tasks, but they often require large, annotated datasets for training. However, collecting such datasets can be time-consuming and costly. Data augmentation techniques provide a solution to this problem by artificially expanding the training dataset. In this study, we propose a deep learning approach for object detection and recognition that leverages data augmentation techniques. We use deep convolutional neural networks (CNNs) as the underlying architecture, specifically focusing on popular models such as You Only Look Once version 3 (YOLOv3). By augmenting the training data with various transformations, such as rotation, scaling, and flipping, we can effectively increase the diversity and size of the dataset. Our approach not only improves the robustness and generalization of the models but also reduces the risk of overfitting. By training on augmented data, the models can learn to recognize objects from different viewpoints, scales, and orientations, leading to improved accuracy and performance. We conduct extensive experiments on benchmark datasets and evaluate the performance of our approach using standard metrics such as precision, recall, and mean average precision (mAP). The experimental results demonstrate that our data augmentation-based deep learning approach achieves superior object detection and recognition accuracy compared to traditional training methods without data augmentation. We compare the average accuracy of the YOLOv3-SPP model with two other variants of the YOLOv3 algorithm: one with a feature extraction network consisting of 53 convolutional layers and the other with 13 convolutional layers. The average accuracy of the proposed model (YOLOv3-SPP) is reported as accuracy of 97%, F1-score of 96%, precision of 94%, and average Intersection over Union (IoU) of 78.04%.

A Noval Approach for Object Recognition Using Decision Tree Clustering by Incorporating Multi-Level BPNN Classifiers and Hybrid Texture Features

This work proposes a novel approach to object recognition, particularly for human faces, based on the principle of human cognition. The suggested approach can handle a dataset or problem with a large number of classes for classification more effectively. The model for the facial recognition-based object detection system was constructed using a combination of decision tree clustering based multi-level Backpropagation neural network classifier-TFMLBPNN-DTC and hybrid texture feature (ILMFD+GLCM) and applied on NS and ORL databases. This model produced the classification accuracy (±standard deviation) of 95.37 ±0.951877% and 90.83 ± 1.374369% for single input and 96.58 ±0.5604582% and 91.50 ± 2.850439% for group-based decision for NS and ORL database respectively. The better classification results encourage its application to other object recognition and classification issues. This work's basic idea also makes it easier to improve classification management for a wide range of classes.

StereoYOLO: A Stereo Vision-Based Method for Maritime Object Recognition and Localization

Image recognition is vital for intelligent ships’ autonomous navigation. However, traditional methods often fail to accurately identify maritime objects’ spatial positions, especially under electromagnetic silence. We introduce the StereoYOLO method, an enhanced stereo vision-based object recognition and localization approach that serves autonomous vessels using only image sensors. It is specifically refined for maritime object recognition and localization scenarios through the integration of convolutional and coordinated attention modules. The method uses stereo cameras to identify and locate maritime objects in images and calculate their relative positions using stereo vision algorithms. Experimental results indicate that the StereoYOLO algorithm boosts the mean Average Precision at IoU threshold of 0.5 (mAP50) in object recognition by 5.23%. Furthermore, the variation in range measurement due to target angle changes is reduced by 6.12%. Additionally, upon measuring the distance to targets at varying ranges, the algorithm achieves an average positioning error of 5.73%, meeting the accuracy and robustness criteria for maritime object collision avoidance on experimental platform ships.

Deep learning approaches for object recognition in plant diseases: a review

Plant diseases pose a significant threat to the economic viability of agriculture and the normal functioning of trees in forests. Accurate detection and identification of plant diseases are crucial for smart agricultural and forestry management. In recent years, the intersection of agriculture and artificial intelligence has become a popular research topic. Researchers have been experimenting with object recognition algorithms, specifically convolutional neural networks, to identify diseases in plant images. The goal is to reduce labor and improve detection efficiency. This article reviews the application of object detection methods for detecting common plant diseases, such as tomato, citrus, maize, and pine trees. It introduces various object detection models, ranging from basic to modern and sophisticated networks, and compares the innovative aspects and drawbacks of commonly used neural network models. Furthermore, the article discusses current challenges in plant disease detection and object detection methods and suggests promising directions for future work in learning-based plant disease detection systems.

Augmented Reality-Assisted Surgical Exposure of an Impacted Tooth: A Pilot Study

Three-dimensional radiological evaluation through cone beam computer tomography is essential in diagnosing and establishing proper surgical management in impacted teeth. Through Augmented Reality (AR), clinicians have the opportunity to use three-dimensional computer-generated radiologic information to visualise the patient and simultaneously the superimposition of his internal structures. Here, we describe a digital workflow to assist the oral surgeon in pre-orthodontic exposure of a vestibular impacted canine using AR. The AR hardware consists of a camera and a traditional stand-up monitor. The registration and tracking are video-based and marker-free, with an automatic pose estimation obtained through VisLab 20.10.1AR software algorithm’s object recognition and tracking approach. A 3D model is created by combining the anterior teeth taken from the intraoral scan with the same teeth plus the included tooth taken from the CBCT segmentation. The 3D file is uploaded into the AR software. Model tracking is straightforward to set up without prior registration of targets or surroundings. The AR information is used successfully to define the surgical access to perform flap and osteotomy. The accuracy of model tracking matching was calculated constantly by the software. During the tracking, the process recorded an inlier ratio of 0.39:0.48. Further studies and clinical trials will evaluate the value of this novel technology in the management of impacted teeth.

Enhancing Human-Robot Collaboration through a Multi-Module Interaction Framework with Sensor Fusion: Object Recognition, Verbal Communication, User of Interest Detection, Gesture and Gaze Recognition.

With the increasing presence of robots in our daily lives, it is crucial to design interaction interfaces that are natural, easy to use and meaningful for robotic tasks. This is important not only to enhance the user experience but also to increase the task reliability by providing supplementary information. Motivated by this, we propose a multi-modal framework consisting of multiple independent modules. These modules take advantage of multiple sensors (e.g., image, sound, depth) and can be used separately or in combination for effective human-robot collaborative interaction. We identified and implemented four key components of an effective human robot collaborative setting, which included determining object location and pose, extracting intricate information from verbal instructions, resolving user(s) of interest (UOI), and gesture recognition and gaze estimation to facilitate the natural and intuitive interactions. The system uses a feature-detector-descriptor approach for object recognition and a homography-based technique for planar pose estimation and a deep multi-task learning model to extract intricate task parameters from verbal communication. The user of interest (UOI) is detected by estimating the facing state and active speakers. The framework also includes gesture detection and gaze estimation modules, which are combined with a verbal instruction component to form structured commands for robotic entities. Experiments were conducted to assess the performance of these interaction interfaces, and the results demonstrated the effectiveness of the approach.

Simple kinesthetic haptics for object recognition

Object recognition is an essential capability when performing various tasks. Humans naturally use either or both visual and tactile perception to extract object class and properties. Typical approaches for robots, however, require complex visual systems or multiple high-density tactile sensors which can be highly expensive. In addition, they usually require actual collection of a large dataset from real objects through direct interaction. In this paper, we propose a kinesthetic-based object recognition method that can be performed with any multi-fingered robotic hand in which the kinematics is known. The method does not require tactile sensors and is based on observing grasps of the objects. We utilize a unique and frame invariant parameterization of grasps to learn instances of object shapes. To train a classifier, training data is generated rapidly and solely in a computational process without interaction with real objects. We then propose and compare between two iterative algorithms that can integrate any trained classifier. The classifiers and algorithms are independent of any particular robot hand and, therefore, can be exerted on various ones. We show in experiments, that with few grasps, the algorithms acquire accurate classification. Furthermore, we show that the object recognition approach is scalable to objects of various sizes. Similarly, a global classifier is trained to identify general geometries (e.g., an ellipsoid or a box) rather than particular ones and demonstrated on a large set of objects. Full scale experiments and analysis are provided to show the performance of the method.

Color object classification using multi-channel Zernike moments-based rotation invariant bag-of-visual-words and deep convolutional neural networks

We propose two rotation-invariant approaches for color object classification and recognition using multi-channel Zernike moments (MZMs)-based bag-of-visual-words (BoVWs) and deep convolutional neural networks (DCNN). The first approach, referred to as MZMs-BoVWs, derives the BoVWs descriptors from a color image and its sub-images after forming its hierarchical sub-divisions and computing the MZMs from each of the sub-images to determine similarity among sub-images to facilitate the construction of BoVWs. The second approach, called DCNN-MZMs, derives the MZMs descriptors from the feature maps of a convolutional layer of a DCNN architecture designed in this paper. A fusion of the descriptors, MZMs-BoVWs and CNN-MZMs, along with the high-performing color histograms (CH) descriptor, i.e., CH+MZMs-BoVWs+CNN-MZMs, is also proposed. The process of the object classification is performed using SVM and its multiple-kernel learning approach. Experimental results show that the proposed approaches outperform the existing state-of-the-art approaches under image-rotation and under limited data for training the CNNs.

SO-YOLOv5: Small object recognition algorithm for sea cucumber in complex seabed environment

Underwater object recognition is a prerequisite for the realization of automated seafood harvesting. To solve the problems of low recognition accuracy and poor generalization ability of existing underwater small object, this paper takes sea cucumber as research objects to research the small object recognition approach in complex underwater environment. In this paper, a sea cucumber dataset is established to solve the problem of lacking sea cucumber dataset in the real seabed environment. A deep learning model SO-YOLOv5 is proposed based on YOLOv5 for underwater small object recognition, which improves the recognition ability of the algorithm for different sizes of objects in complex environment. The proposed model embeds a large-scale feature extraction layer in structure to increase the detection ability of small objects by referring to the idea of the Bi-directional Feature Pyramid Network. SO-YOLOv5 fuses the features between deep and shallow layers and balances the feature information of different scales, and integrates the Coordinate Attention mechanism to enhance the sensitivity of the algorithm to the direction and position information. The experimental results illustrate that the mAP of the proposed approach achieves 95.47% for sea cucumber recognition in the complex seabed environment, which is 3.42%, 6.79%, and 5.46% higher than the traditional Faster R-CNN, SSD, and YOLOv5 algorithms, respectively. This research not only provides an effective approach for sea cucumber recognition in complex environment but also has certain reference significance for the recognition of small objects in other complex environments. In addition, the proposed approach has practical application value to improve intelligence level in aquaculture.

A Generative Deep Learning Approach for Shape Recognition of Arbitrary Objects from Phaseless Acoustic Scattering Data

A generative deep learning approach for shape recognition of an arbitrary object from its acoustic scattering properties is proposed and demonstrated. The strategy exploits deep neural networks to learn the mapping between the latent space of a 2D acoustic object and the far‐field scattering amplitudes. A neural network is designed as an adversarial autoencoder and trained via unsupervised learning to determine the latent space of the acoustic object. Important structural features of the object are embedded in lower‐dimensional latent space which supports the modeling of a shape generator and accelerates the learning in the inverse design process. The proposed inverse design uses the variational inference approach with encoder‐ and decoder‐like architecture where the decoder is composed of two pretrained neural networks: the generator and the forward model. The data‐driven framework finds an accurate solution to the ill‐posed inverse scattering problem, where nonunique solution space is overcome by the multifrequency phaseless far‐field patterns. This inverse method is a powerful design tool that doesn't require complex analytical calculation and opens up new avenues for practical realization, automatic recognition of arbitrary‐shaped submarines or large fish, and other underwater applications.

A Shape-based Approach for Recognition of Hidden Objects Using Microwave Radar Imaging System

A Shape-based Approach for Recognition of Hidden Objects Using Microwave Radar Imaging System

A review of deep learning based approach for fruit object detection and recognition

A review of deep learning based approach for fruit object detection and recognition

Semi-supervised learning approach for localization and pose estimation of texture-less objects in cluttered scenes

3D object recognition and 6D pose estimation are crucial and fundamental endeavours for industrial assembly line automation such as robotic controlled pick-and-place. While the problem on textured objects is extensively studied, it is still an open research topic for texture-less industrial parts, e.g, solid cylinder and hollow tube, which are symmetric and appear similar in shapes from many viewing perspectives, causing pose ambiguity. Also, the industrial assembly line environment is usually cluttered and the captured data is noisy, which makes this task even more challenging. In this paper, we propose a novel object localization and pose estimation technique using RGB images and depth maps of industrial assembly parts. Our segmentation model is fully morphological and unsupervised for localizing the region of interest containing the target object extracted from the depth map. Our segmentation technique is effective in the presence of partial occlusion, multiple objects, and cluttered scenes. We use a model based approach for object recognition based on Stochastic Gradient Descent trained on features of Histogram of Oriented Gradients (HOG) and invariant moments of the region of interest containing the target object. We generate synthetic training images automatically from the CAD models of the industrial parts. We use a contour matching strategy based on Dynamic Time Warping (DTW) algorithm to estimate the optimal 6D pose of the object from a set of candidates. Experimental results show that our proposed approach competes and demonstrates advantages on the challenging T-LESS dataset compared to the state-of-the-art methods. • We introduce a strategy for generating multiple viewpoint synthetic images from a CAD model. • We introduce a non-supervised clustering process for object segmentation based on a depth map that does not require a training stage. • We demonstrate how different HOG feature parameters can be tuned to achieve better prediction and performance. • We present a new real-time strategy for object pose estimation using Dynamic Time Warping (DTW) between the target object contour and best prior candidates. • We show that our method achieves better time performance than other works for pose estimation.

3D Object Recognition Using Fast Overlapped Block Processing Technique

Three-dimensional (3D) image and medical image processing, which are considered big data analysis, have attracted significant attention during the last few years. To this end, efficient 3D object recognition techniques could be beneficial to such image and medical image processing. However, to date, most of the proposed methods for 3D object recognition experience major challenges in terms of high computational complexity. This is attributed to the fact that the computational complexity and execution time are increased when the dimensions of the object are increased, which is the case in 3D object recognition. Therefore, finding an efficient method for obtaining high recognition accuracy with low computational complexity is essential. To this end, this paper presents an efficient method for 3D object recognition with low computational complexity. Specifically, the proposed method uses a fast overlapped technique, which deals with higher-order polynomials and high-dimensional objects. The fast overlapped block-processing algorithm reduces the computational complexity of feature extraction. This paper also exploits Charlier polynomials and their moments along with support vector machine (SVM). The evaluation of the presented method is carried out using a well-known dataset, the McGill benchmark dataset. Besides, comparisons are performed with existing 3D object recognition methods. The results show that the proposed 3D object recognition approach achieves high recognition rates under different noisy environments. Furthermore, the results show that the presented method has the potential to mitigate noise distortion and outperforms existing methods in terms of computation time under noise-free and different noisy environments.

Object recognition in medical images via anatomy-guided deep learning.

Despite advances in deep learning, robust medical image segmentation in the presence of artifacts, pathology, and other imaging shortcomings has remained a challenge. In this paper, we demonstrate that by synergistically marrying the unmatched strengths of high-level human knowledge (i.e., natural intelligence (NI)) with the capabilities of deep learning (DL) networks (i.e., artificial intelligence (AI)) in garnering intricate details, these challenges can be significantly overcome. Focusing on the object recognition task, we formulate an anatomy-guided deep learning object recognition approach named AAR-DL which combines an advanced anatomy-modeling strategy, model-based non-deep-learning object recognition, and deep learning object detection networks to achieve expert human-like performance. The AAR-DL approach consists of 4 key modules wherein prior knowledge (NI) is made use of judiciously at every stage. In the first module AAR-R, objects are recognized based on a previously created fuzzy anatomy model of the body region with all its organs following the automatic anatomy recognition (AAR) approach wherein high-level human anatomic knowledge is precisely codified. This module is purely model-based with no DL involvement. Although the AAR-R operation lacks accuracy, it is robust to artifacts and deviations (much like NI), and provides the much-needed anatomic guidance in the form of rough regions-of-interest (ROIs) for the following DL modules. The 2nd module DL-R makes use of the ROI information to limit the search region to just where each object is most likely to reside and performs DL-based detection of the 2D bounding boxes (BBs) in slices. The 2D BBs hug the shape of the 3D object much better than 3D BBs and their detection is feasible only due to anatomy guidance from AAR-R. In the 3rd module, the AAR model is deformed via the found 2D BBs providing refined model information which now embodies both NI and AI decisions. The refined AAR model more actively guides the 4th refined DL-R module to perform final object detection via DL. Anatomy knowledge is made use of in designing the DL networks wherein spatially sparse objects and non-sparse objects are handled differently to provide the required level of attention for each. Utilizing 150 thoracic and 225 head and neck (H&N) computed tomography (CT) data sets of cancer patients undergoing routine radiation therapy planning, the recognition performance of the AAR-DL approach is evaluated on 10 thoracic and 16 H&N organs in comparison to pure model-based approach (AAR-R) and pure DL approach without anatomy guidance. Recognition accuracy is assessed via location error/ centroid distance error, scale or size error, and wall distance error. The results demonstrate how the errors are gradually and systematically reduced from the 1st module to the 4th module as high-level knowledge is infused via NI at various stages into the processing pipeline. This improvement is especially dramatic for sparse and artifact-prone challenging objects, achieving a location error over all objects of 4.4mm and 4.3mm for the two body regions, respectively. The pure DL approach failed on several very challenging sparse objects while AAR-DL achieved accurate recognition, almost matching human performance, showing the importance of anatomy guidance for robust operation. Anatomy guidance also reduces the time required for training DL networks considerably. (i) High-level anatomy guidance improves recognition performance of DL methods. (ii) This improvement is especially noteworthy for spatially sparse, low-contrast, inconspicuous, and artifact-prone objects. (iii) Once anatomy guidance is provided, 3D objects can be detected much more accurately via 2D BBs than 3D BBs and the 2D BBs represent object containment with much more specificity. (iv) Anatomy guidance brings stability and robustness to DL approaches for object localization. (v) The training time can be greatly reduced by making use of anatomy guidance.