Object Identification Task Research Articles

Harnessing slow event-related fMRI to investigate trial-level brain-behavior relationships during object identification.

Understanding brain-behavior relationships is the core goal of cognitive neuroscience. However, these relationships-especially those related to complex cognitive and psychopathological behaviors-have recently been shown to suffer from very small effect sizes (0.1 or less), requiring potentially thousands of participants to yield robust findings. Here, we focus on a much more optimistic case utilizing task-based fMRI and a multi-echo acquisition with trial-level brain-behavior associations measured within participant. In a visual object identification task for which the behavioral measure is response time (RT), we show that while trial-level associations between BOLD and RT can similarly suffer from weak effect sizes, converting these associations to their corresponding group-level effects can yield robust peak effect sizes (Cohen's d = 1.0 or larger). Multi-echo denoising (Multi-Echo ICA or ME-ICA) yields larger effects than optimally combined multi-echo with no denoising, which is in turn an improvement over standard single-echo acquisition. While estimating these brain-behavior relationships benefits from the inclusion of a large number of trials per participant, even a modest number of trials (20-30 or more) yields robust group-level effect sizes, with replicable effects obtainable with relatively standard sample sizes (N = 20-30 participants per sample).

Stereoscopic depth perception through foliage

Both humans and computational methods struggle to discriminate the depths of objects hidden beneath foliage. However, such discrimination becomes feasible when we combine computational optical synthetic aperture sensing with the human ability to fuse stereoscopic images. For object identification tasks, as required in search and rescue, wildlife observation, surveillance, and early wildfire detection, depth assists in differentiating true from false findings, such as people, animals, or vehicles vs. sun-heated patches at the ground level or in the tree crowns, or ground fires vs. tree trunks. We used video captured by a drone above dense woodland to test users’ ability to discriminate depth. We found that this is impossible when viewing monoscopic video and relying on motion parallax. The same was true with stereoscopic video because of the occlusions caused by foliage. However, when synthetic aperture sensing was used to reduce occlusions and disparity-scaled stereoscopic video was presented, whereas computational (stereoscopic matching) methods were unsuccessful, human observers successfully discriminated depth. This shows the potential of systems which exploit the synergy between computational methods and human vision to perform tasks that neither can perform alone.

Object Detection Using Capsule Neural Network: An Overview

Appropriate remedies for tasks like language translation, object identification, object segmentation, picture recognition, and natural language processing are needed for today's computer vision tasks. This paper explores the use of Capsule neural Networks (Caps Nets) in object identification and offers a thorough analysis of their developments and uses. It presents an analysis of the transformative impact of Caps Nets on object identification tasks. This distinctive the architecture of neural network is presented as an alternative to convolutional neural networks (CNNs). The paper highlights that how Caps Nets are unique in capturing spatial correlations and hierarchical patterns in visualization data by analyzing the fundamental ideas of the technology. Additionally, it demonstrates how Caps net out perform CNNs in terms of generalization, interpretability, and last in the resistance to spatial distortions, confirming their goodness at object detection. By integrating the Caps networks with the latest scientific findings and advances, the paper shows the current status and potential future paths for object detection methods that use this leading neural network architecture.

Object Detection for Unmanned Aerial Vehicles: A Comprehensive Review

Goal: Researchers studying artificial intelligence have focused a lot of emphasis on computer vision in drones. Drones with intelligence can tackle a lot of issues in real time. For the purpose of monitoring particular surroundings, computer vision tasks like object identification, object tracking, and object counting are important. It becomes increasingly difficult to do, though, due to elements like motion blur, occlusion, camera angle, and altitude. Methodology: A thorough assessment of the literature on object identification and tracking with unmanned aerial vehicles (UAVs) in relation to various applications has been done for this research. This study highlights the research gaps and provides a summary of the results of previous studies. Contribution: Detailed and categorized object identification techniques are used in UAV photos. A selection of UAV datasets tailored to object identification tasks is provided. Summaries of current research projects in various applications are provided. In order to alleviate highlighted research limitations, a secure onboard processing system on a strong object detection framework in precision agriculture is finally presented.

Visual image design of the internet of things based on AI intelligence

Visual object detection has emerged as a critical technology for Unmanned Arial Vehicle (UAV) use due to advances in computer vision. New developments in fields like communication technology and the UAV needs to be able to act autonomously by gathering data and then making choices. These tendencies have brought us to cutting-edge levels of health care, transportation, energy, monitoring, and security for visual image detection and manufacturing endeavors. These include coordination in communication via IoT, sustainability of IoT network, and optimization challenges in path planning. Because of their limited battery life, these gadgets are limited in their range of communication. UAVs can be seen as terminal devices connected to a large network where a swarm of other UAVs is coordinating their motions, directing one another, and maintaining watch over locations outside its visual range. One of the essential components of UAV-based applications is the ability to recognize objects of interest in aerial photographs taken by UAVs. While aerial photos might be useful, object detection is challenging. As a result, capturing aerial photographs with UAVs is a unique challenge since the size of things in these images might vary greatly. The study proposal included specific information regarding the Detection of Visual Images by UAVs (DVI-UAV) using the IoT and Artificial Intelligence (AI). Included in the study of AI is the concept of DSYolov3. The DSYolov3 model was presented to deal with these problems in the UAV industry. By fusing the channel-wise feature across multiple scales using a spatial pyramid pooling approach, the proposed study creates a novel module, Multi-scale Fusion of Channel Attention (MFCAM), for scale-variant object identification tasks. The method's effectiveness and efficiency have been thoroughly tested and evaluated experimentally. The suggested method would allow us to outperform most current detectors and guarantee that the models will be useable on UAVs. There will be a 95 % success rate in terms of visual image detection, a 94 % success rate in terms of computation cost, a 97 % success rate in terms of accuracy, and a 95 % success rate in terms of effectiveness.

A Comprehensive Review on Multiple Instance Learning

Multiple-instance learning has become popular over recent years due to its use in some special scenarios. It is basically a type of weakly supervised learning where the learning dataset contains bags of instances instead of a single feature vector. Each bag is associated with a single label. This type of learning is flexible and a natural fit for multiple real-world problems. MIL has been employed to deal with a number of challenges, including object detection and identification tasks, content-based image retrieval, and computer-aided diagnosis. Medical image analysis and drug activity prediction have been the main uses of MIL in biomedical research. Many Algorithms based on MIL have been put forth over the years. In this paper, we will discuss MIL, the background of MIL and its application in multiple domains, some MIL-based methods, challenges, and lastly, the conclusions and prospects.

Artificial vision techniques at the frontiers of video surveillance

The present work addresses the task of identifying a predatory behavior of robbery of homes or businesses. The proposed objective is the detection of blunt elements used in the commission of the crime, limiting the context to barrettes, covered faces, people and gates (doors or windows). The proposal addresses the task of object identification applying Single Shot Detectors (SSD). Due to its versatility and the physical resources applied, the structure of SSD ResNet50 V1 FPN 640x640 has been chosen from the TensorFlow Model Zoo to train and validate the classification. This has been built in five classes, for the training and validation set, an average of 50 annotations per class have been processed. Additionally, a support function was worked on in the detection of human activity. The evaluated model obtained a mA of 69% in the detection of objects and in the identification of criminal behavior it showed a performance of 69%.

Animal Species Recognition with Deep Convolutional Neural Networks from Ecological Camera Trap Images.

Accurate identification of animal species is necessary to understand biodiversity richness, monitor endangered species, and study the impact of climate change on species distribution within a specific region. Camera traps represent a passive monitoring technique that generates millions of ecological images. The vast numbers of images drive automated ecological analysis as essential, given that manual assessment of large datasets is laborious, time-consuming, and expensive. Deep learning networks have been advanced in the last few years to solve object and species identification tasks in the computer vision domain, providing state-of-the-art results. In our work, we trained and tested machine learning models to classify three animal groups (snakes, lizards, and toads) from camera trap images. We experimented with two pretrained models, VGG16 and ResNet50, and a self-trained convolutional neural network (CNN-1) with varying CNN layers and augmentation parameters. For multiclassification, CNN-1 achieved 72% accuracy, whereas VGG16 reached 87%, and ResNet50 attained 86% accuracy. These results demonstrate that the transfer learning approach outperforms the self-trained model performance. The models showed promising results in identifying species, especially those with challenging body sizes and vegetation.

Perception of global properties, objects, and settings in natural auditory scenes

Theories of auditory scene analysis suggest our perception of scenes relies on identifying and segregating objects within it. However, a more global process may occur while analyzing scenes, which has been evidenced in the visual domain. In our first experiment, we studied perception of eight global properties (e.g., openness), using a collection of 200 high-quality auditory scenes. Participants showed high agreement on their ratings of global properties. The global properties were explained by a two-factor model. Acoustic features of scenes were explained by a seven-factor model, and linearly predicted the global ratings by different amounts (R-squared = 0.33–0.87), although we also observed nonlinear relationships between acoustical and global variables. A multi-layer neural network trained to recognize auditory objects in everyday soundscapes from YouTube shows high-level embeddings of our 200 scenes are correlated with some global variables at earlier stages of processing than others. In a second experiment, we evaluated participants’ accuracy in identifying the setting of and objects within scenes across three durations (1, 2, and 4 s). Overall, participants performed better on the object identification task, but needed longer duration stimuli to do so. These results suggest object identification may require more processing time and/or attention switching than setting identification.

Development of object identification model with deep reinforcement learning algorithm

This research work presents an object identification method based on the machine learning technique based on human vision system. The objective is to prevent processing a complete image in sort to locate objects. Presently, the-state-of-the-art techniques divide an image into sub-regions and search for an object in all the subparts. This is ineffective for applications like embedded systems where the computation power is restricted or the resolution of the images are high. To address this issue, an object identification task was formulated as a decision-making problem. Followed the concept of DRL proposed, accepted RL algorithm DQL was applied to learn a policy from input data, i.e. images, to identify objects in a scene. In this manner, with the policy learned, a set of actions that transforms a box was apply in order to make tighter a bounding box around the target object.

Ventral and Dorsal Stream EEG Channels: Key Features for EEG-Based Object Recognition and Identification.

Object recognition and object identification are multifaceted cognitive operations that require various brain regions to synthesize and process information. Prior research has evidenced the activity of both visual and temporal cortices during these tasks. Notwithstanding their similarities, object recognition and identification are recognized as separate brain functions. Drawing from the two-stream hypothesis, our investigation aims to understand whether the channels within the ventral and dorsal streams contain pertinent information for effective model learning regarding object recognition and identification tasks. By utilizing the data we collected during the object recognition and identification experiment, we scrutinized EEGNet models, trained using channels that replicate the two-stream hypothesis pathways, against a model trained using all available channels. The outcomes reveal that the model trained solely using the temporal region delivered a high accuracy level in classifying four distinct object categories. Specifically, the object recognition and object identification models achieved an accuracy of 89% and 85%, respectively. By incorporating the channels that mimic the ventral stream, the model's accuracy was further improved, with the object recognition model and object identification model achieving an accuracy of 95% and 94%, respectively. Furthermore, the Grad-CAM result of the trained models revealed a significant contribution from the ventral and dorsal stream channels toward the training of the EEGNet model. The aim of our study is to pinpoint the optimal channel configuration that provides a swift and accurate brain-computer interface system for object recognition and identification.

The Moderating Effects of Task Complexity and Age on the Relationship between Automation Use and Cognitive Workload

In recent decades, automation has become increasingly integrated into knowledge work environments, i.e., jobs that require specialized skills for task completion. Some research has shown that automation may reduce cognitive workload. However, other work reports that increasing automation does not always lower cognitive workload. These conflicting results suggest that further investigation is needed to identify factors that influence the relationship between automation use and cognitive workload. Therefore, this study aimed to investigate how two potential moderators, task complexity and age, influence that relationship. A total of 24 younger and 24 middle-aged adults performed an object identification task that resembled a baggage scanning procedure conducted by airport security personnel. The task was manipulated by varying the level of automation (4 increasing levels) that provided support in identifying objects of interest as well as the complexity of task (two types: component and coordinative). In addition, participants completed a pattern identification and sorting memory task, which helped emulate a knowledge work environment. Performance (i.e., object identification accuracy, average completion time, pattern identification accuracy, and sorting memory task accuracy) and subjective workload (NASA-RTLX score) measures were recorded. Participants also rated the usability of the automation. Overall, results showed that while the use of automation was associated with reduced cognitive workload, both types of task complexity negatively affected this relationship such that increased complexity was associated with a decrease in accuracy. Age did not moderate the relationship between automation and cognitive workload, but there were qualitative differences in terms of how the two age groups perceived the utility and usability of the automated systems. Knowledge generated from this research has implications for the design of future human-automation systems and can be used to inform interface design that is tailored to the needs and preferences of different users and use cases.

The pupillometry of the possible: an investigation of infants' representation of alternative possibilities.

Contrasting possibilities has a fundamental adaptive value for prediction and learning. Developmental research, however, has yielded controversial findings. Some data suggest that preschoolers might have trouble in planning actions that take into account mutually exclusive possibilities, while other studies revealed an early understanding of alternative future outcomes based on infants' looking behaviour. To better understand the origin of such abilities, here we use pupil dilation as a potential indicator of infants' representation of possibilities. Ten- and 14-month-olds were engaged in an object-identification task by watching video animations where three different objects with identical top parts moved behind two screens. Importantly, a target object emerged from one of the screens but remained in partial occlusion, revealing only its top part, which was compatible with a varying number of possible identities. Just as adults' pupil diameter grows monotonically with the amount of information held in memory, we expected that infants' pupil size would increase with the number of alternatives sustained in memory as candidate identities for the partially occluded object. We found that pupil diameter increased with the object's potential identities in 14- but not in 10-month-olds. We discuss the implications of these results for the foundation of humans' capacities to represent alternatives. This article is part of the theme issue 'Thinking about possibilities: mechanisms, ontogeny, functions and phylogeny'.

YOLO-LRDD: a lightweight method for road damage detection based on improved YOLOv5s

In computer vision, timely and accurate execution of object identification tasks is critical. However, present road damage detection approaches based on deep learning suffer from complex models and computationally time-consuming issues. To address these issues, we present a lightweight model for road damage identification by enhancing the YOLOv5s approach. The resulting algorithm, YOLO-LRDD, provides a good balance of detection precision and speed. First, we propose the novel backbone network Shuffle-ECANet by adding an ECA attention module into the lightweight model ShuffleNetV2. Second, to ensure reliable detection, we employ BiFPN rather than the original feature pyramid network since it improves the network's capacity to describe features. Moreover, in the model training phase, localization loss is modified to Focal-EIOU in order to get higher-quality anchor box. Lastly, we augment the well-known RDD2020 dataset with many samples of Chinese road scenes and compare YOLO-LRDD against several state-of-the-art object detection techniques. The smaller model of our YOLO-LRDD offers superior performance in terms of accuracy and efficiency, as determined by our experiments. Compared to YOLOv5s in particular, YOLO-LRDD improves single image recognition speed by 22.3% and reduces model size by 28.8% while maintaining comparable accuracy. In addition, it is easier to implant in mobile devices because its model is smaller and lighter than those of the other approaches.

Perceptual–Cognitive Expertise in Law Enforcement: An Object-Identification Task

The few perceptual–cognitive expertise and deception studies in the domain of law enforcement have yet to examine perceptual–cognitive expertise differences of police trainees and police officers. The current study uses methods from the perceptual–cognitive expertise and deception models. Participants watched temporally occluded videos of actors honestly drawing a weapon and deceptively drawing a non-weapon from a concealed location on their body. Participants determined if the actor was holding a weapon or a non-weapon. Using signal-detection metrics—sensitivity and response bias—we did not find evidence of perceptual–cognitive expertise; performance measures did not differ significantly between police trainees and experienced officers. However, consistent with the hypotheses, we did find that both police trainees and police officers became more sensitive in identifying the object as occlusion points progressed. Additionally, we found that across police trainees and police officers, their response bias became more liberal (i.e., more likely to identify the object as a weapon) as occlusion points progressed. This information has potential impacts for law enforcement practices and additional research.

Comparison of Object Detection in Head-Mounted and Desktop Displays for Congruent and Incongruent Environments

Virtual reality technologies, including head-mounted displays (HMD), can provide benefits to psychological research by combining high degrees of experimental control with improved ecological validity. This is due to the strong feeling of being in the displayed environment (presence) experienced by VR users. As of yet, it is not fully explored how using HMDs impacts basic perceptual tasks, such as object perception. In traditional display setups, the congruency between background environment and object category has been shown to impact response times in object perception tasks. In this study, we investigated whether this well-established effect is comparable when using desktop and HMD devices. In the study, 21 participants used both desktop and HMD setups to perform an object identification task and, subsequently, their subjective presence while experiencing two-distinct virtual environments (a beach and a home environment) was evaluated. Participants were quicker to identify objects in the HMD condition, independent of object-environment congruency, while congruency effects were not impacted. Furthermore, participants reported significantly higher presence in the HMD condition.

Intracortical microstimulation of somatosensory cortex enables object identification through perceived sensations.

Advances in brain-machine interfaces have helped restore function and independence for individuals with sensorimotor deficits; however, providing efficient and effective sensory feedback remains challenging. Intracortical microstimulation (ICMS) of sensorimotor brain regions is a promising technique for providing bioinspired sensory feedback. In a human participant with chronically-implanted microelectrode arrays, we provided ICMS to the primary somatosensory cortex to generate tactile percepts in his hand. In a 3-choice object identification task, the participant identified virtual objects using tactile sensory feedback and no visual information. We evaluated three different stimulation paradigms, each with a different weighting of the grip force and its derivative, to explore the potential benefits of a more bioinspired stimulation strategy. In all paradigms, the participant's ability to identify the objects was above-chance, with object identification accuracy reaching 80% correct when using only sustained grip force feedback and 76.7% when using equal weighting of both sustained grip force and its derivative. These results demonstrate that bioinspired ICMS can provide sensory feedback that is functionally beneficial in sensorimotor tasks. Designing more efficient stimulation paradigms is important because it will allow us to 1) provide safer stimulation delivery methods that reduce overall injected charge without sacrificing function and 2) more effectively transmit sensory information to promote intuitive integration and usage by the human body.

Dog breed classification using convolution neural network

At present, there are numerous applications of machine learning that have been identified. The identification and recognition of the object is very popular and significant among them. This paper presents a model for dog breed classification using convolution neural network (CNN) that is a very popular model for image processing. In this work, a variant CNN model is used to classify different breeds of dogs which uses advanced filters to detect the patterns that exists within a given image and fixing randomised filter on an image to figure out the available patterns. The proposed model is constructed to calibrate the weights by associating data in front of the Xception model. The object identification task gets superhuman potential in the Xception model and it is trained over 15 millions of good resolution images with around 22,000 different object classifications. The experimental result reveals that the proposed model performs well in terms of accuracy, i.e., 98%.

Visual texture agnosia influences object identification in dementia with Lewy bodies and Alzheimer's disease

BackgroundNumerous studies have shown visuoperceptual/visuospatial deficits in dementia with Lewy bodies (DLB) and Alzheimer's disease (AD). Visual texture recognition is also impaired in patients with DLB and AD. Although patients with DLB often exhibit visual misidentifications of objects, there are few studies on the relationships between visual texture recognition and viewpoints for object recognition. ObjectivesThe aim of this study was to clarify how viewpoints, textures, and visual cognitive functions affect object recognition and result in visual misidentifications in patients with DLB or AD. MethodsA total of 37 patients with probable DLB and 58 with probable AD and 32 age-matched healthy controls underwent neuropsychological and visuoperceptual assessments, and performed object identification tasks under four conditions (non-canonical view + blurry texture, non-canonical view + clear texture, canonical view + blurry texture, canonical view + clear texture). The relationship between object identification and other visuoperceptual functions was analyzed. ResultsPatients with DLB and AD exhibited significantly impaired object recognition under non-canonical viewing with blurry texture conditions, with the DLB patients exhibiting a significantly worse performance than the AD patients. Patients with DLB and AD exhibited visual misidentifications during object identification tasks under non-canonical viewing. In patients with DLB, the number of visual misidentifications was significantly correlated with the scores of visual texture recognition. ConclusionsThe present study showed that significantly impaired object recognition in patients with DLB under the influences by both viewpoint and visual texture and in those with AD under the influence by viewpoint. Visual misidentification in object recognition could be associated with impaired visual texture recognition in DLB.

Learning efficient haptic shape exploration with a rigid tactile sensor array.

Haptic exploration is a key skill for both robots and humans to discriminate and handle unknown objects or to recognize familiar objects. Its active nature is evident in humans who from early on reliably acquire sophisticated sensory-motor capabilities for active exploratory touch and directed manual exploration that associates surfaces and object properties with their spatial locations. This is in stark contrast to robotics. In this field, the relative lack of good real-world interaction models—along with very restricted sensors and a scarcity of suitable training data to leverage machine learning methods—has so far rendered haptic exploration a largely underdeveloped skill. In robot vision however, deep learning approaches and an abundance of available training data have triggered huge advances. In the present work, we connect recent advances in recurrent models of visual attention with previous insights about the organisation of human haptic search behavior, exploratory procedures and haptic glances for a novel architecture that learns a generative model of haptic exploration in a simulated three-dimensional environment. This environment contains a set of rigid static objects representing a selection of one-dimensional local shape features embedded in a 3D space: an edge, a flat and a convex surface. The proposed algorithm simultaneously optimizes main perception-action loop components: feature extraction, integration of features over time, and the control strategy, while continuously acquiring data online. Inspired by the Recurrent Attention Model, we formalize the target task of haptic object identification in a reinforcement learning framework and reward the learner in the case of success only. We perform a multi-module neural network training, including a feature extractor and a recurrent neural network module aiding pose control for storing and combining sequential sensory data. The resulting haptic meta-controller for the rigid 16 × 16 tactile sensor array moving in a physics-driven simulation environment, called the Haptic Attention Model, performs a sequence of haptic glances, and outputs corresponding force measurements. The resulting method has been successfully tested with four different objects. It achieved results close to 100% while performing object contour exploration that has been optimized for its own sensor morphology.