Eye Region Research Articles

Gaze Zone Classification for Driving Studies Using YOLOv8 Image Classification

Gaze zone detection involves estimating where drivers look in terms of broad categories (e.g., left mirror, speedometer, rear mirror). We here specifically focus on the automatic annotation of gaze zones in the context of road safety research, where the system can be tuned to specific drivers and driving conditions, so that an easy to use but accurate system may be obtained. We show with an existing dataset of eye region crops (nine gaze zones) and two newly collected datasets (12 and 10 gaze zones) that image classification with YOLOv8, which has a simple command line interface, achieves near-perfect accuracy without any pre-processing of the images, as long as a model is trained on the driver and conditions for which annotation is required (such as whether the drivers wear glasses or sunglasses). We also present two apps to collect the training images and to train and apply the YOLOv8 models. Future research will need to explore how well the method extends to real driving conditions, which may be more variable and more difficult to annotate for ground truth labels.

Assessment of CCMP in Capturing High Winds with Respect to Individual Satellite Datasets

High-wind structures were identified in the Cross-Calibrated Multi-Platform (CCMP) ocean wind vector reanalysis for comparison with winds measured by satellite radiometers, scatterometers, and synthetic aperture radar (SAR) instruments from February to October 2023. The comparison aims to evaluate bias, uncertainty, and spatial correlations with the goal of enhancing the accuracy of ocean wind datasets during tropical cyclones (TCs). In 10° longitude × 10° latitude blocks, each containing a TC, Soil Moisture Active Passive (SMAP) and Advanced Microwave Scanning Radiometer 2 (AMSR2) winds are 6.5 and 4.8% higher than CCMP, while Advanced Scatterometer (ASCATB) is 0.8% lower. For extratropical cyclones, AMSR2 and SMAP also show stronger winds with a 5% difference, and ASCATB is about 0.3% weaker compared to CCMP. The comparison between SAR and CCMP for TC winds, sampled at the locations and time frames of SAR tiles, indicates that SAR winds around TCs are about 9% higher than CCMP winds. Using empirically defined TC structural indices, we find that the TCs observed by CCMP are shifted in locations and lack a compact core region. A Random Forest (RF) regressor was applied to TCs in CCMP with corresponding SAR observations, nearly correcting the full magnitude of low bias in CCMP statistically, with a 15 m/s correction in the core region. The hierarchy of importance among the predictors is as follows: CCMP wind speed (62%), distance of SAR pixels to the eye region (21%) and eye center (7%), and distance of CCMP pixels to the eye region (5%) and eye center (5%).

VITILIGO TREATMENT WITH MANIPULATED FORMULATION CONTAINING TOPICAL ANTI-INFLAMMATORY DRUGS: CASE REPORT

Introduction: Vitiligo is a dermatological condition that causes white patches on the skin due to the loss of melanocytes. Classified as an autoimmune disease, it is associated with genetic, environmental factors and oxidative stress. It occurs predominantly in people aged between 10 and 30 years, affecting the self-esteem and quality of life of patients. Objective: This study aimed to monitor a patient using a compounded formulation containing topical anti-inflammatories, regarding improvement in the pigmentation of the spots. Case Report: The volunteer presented vitiligo spots in the eye and eyelid region, leading to self-esteem problems because they are quite visible spots. The formulation containing topical anti-inflammatories was compounded in a Compounding Pharmacy in the city of Maringá (PR) and given to the volunteer, with instructions for use to apply once a day directly on the spots, always at night, after cleaning the affected area. The monitoring was carried out by the students weekly, through photos and application of a questionnaire to the volunteer. The treatment was carried out for 30 days, without any complications, discomfort or skin irritation. The patient reported having undergone the treatment without interruption, and after the total treatment time, there was a significant improvement in pigmentation. In the questionnaire applied, the volunteer reported that there was an improvement in the appearance of the spots, and the result can be seen in the weekly photos. This research was approved by the Ethics Committee for Research involving Human Beings of Uningá (CEP/Uningá), under protocol number 6,950,683.

Gaze behavior in response to affect during natural social interactions.

Attention in social interactions is directed by social cues such as the face or eye region of an interaction partner. Several factors that influence these attentional biases have been identified in the past. However, most findings are based on paradigms with static stimuli and no interaction potential. Therefore, the current study investigated the influence of one of these factors, namely facial affect in natural social interactions using an evaluated eye-tracking setup. In a sample of 35 female participants, we examined how individuals' gaze behavior responds to changes in the facial affect of an interaction partner trained in affect modulation. Our goal was to analyze the effects on attention to facial features and to investigate their temporal dynamics in a natural social interaction. The study results, obtained from both aggregated and dynamic analyses, indicate that facial affect has only subtle influences on gaze behavior during social interactions. In a sample with high measurement precision, these findings highlight the difficulties of capturing the subtleties of social attention in more naturalistic settings. The methodology used in this study serves as a foundation for future research on social attention differences in more ecologically valid scenarios.

Glial Cell Responses and Gene Expression Dynamics in Retinas of Treated and Untreated RPE65 Mutant Dogs.

The long-term evaluation of RPE65 gene augmentation initiated in middle-aged RPE65 mutant dogs previously uncovered notable inter-animal and intra-retinal variations in treatment efficacy. The study aims to gain deeper insights into the status of mutant retinas and assess the treatment impact. Immunohistochemistry utilizing cell-specific markers and reverse transcription-quantitative PCR (RT-qPCR) analysis were conducted on archival retinal sections from normal and RPE65 mutant dogs. Untreated middle-aged mutant retinas exhibited marked downregulation in the majority of 20 examined genes associated with key retinal pathways. These changes were accompanied by a moderate increase in microglia numbers, altered expression patterns of glial-neuronal transmitter recycling proteins, and gliotic responses in Müller glia. Analysis of advanced-aged mutant dogs revealed mild outer nuclear layer loss in the treated eye compared to moderate loss in the corresponding retinal regions of the untreated control eye. However, persistent Müller glial stress response along with photoreceptor synapse loss were evident in both treated and untreated eyes. Photoreceptor synaptic remodeling, infrequent in treated regions, was observed in all untreated advanced-aged retinas, accompanied by a progressive increase in microglial cells indicative of ongoing inflammation. Interestingly, about half of the examined genes showed similar expression levels between treated and untreated advanced-aged mutant retinas, with some reaching normal levels. Gene expression data suggest a shift from pro-degenerative mechanisms in middle-aged mutant retinas to more compensatory mechanisms in preserved retinal regions at advanced stages, despite ongoing degeneration. Such shift, potentially attributed to a number of surviving resilient cells, may influence disease patterns and treatment outcomes.

Hierarchical Event-RGB Interaction Network for single-eye expression recognition

The Single-eye Expression Recognition task stands as a crucial vision task, aimed at decoding human emotional states through careful examination of the eye region. Nevertheless, traditional cameras face challenges in detecting and capturing relevant biological information, especially under demanding lighting conditions such as dim environments, high exposure scenarios, or when other radiation sources are present. In this regard, we use a new type of sensor data that can resist extreme lighting conditions, namely event camera data, to improve the performance of single-eye expression recognition. To this end, we propose a novel Hierarchical Event-RGB Interaction Network (HI-Net), to fully integrate RGB and event data to overcome the extreme lighting challenges faced by the single-eye expression recognition task. The HI-Net contains two novel designs: Event-RGB Semantic Interaction Mechanism (ER-SIM) and Hierarchical Semantics Modeling (HSM) Scheme. The former aims to achieve interaction between Event and RGB modality semantics, while the latter aims to obtain high-quality modality semantic representations. In the ER-SIM, we employ an effective cross-attention mechanism to facilitate information fusion, to adaptively integrate and complement multi-scale Event and RGB semantics to cope with extreme lighting conditions. In HSM Scheme, we first explore multi-scale contextual semantics for the event modality and the RGB modality respectively. Then, we perform a semantics interaction strategy for these multi-scale contextual semantics, to enhance each modality's semantic representation. Extensive experiments demonstrate that our HI-Net significantly outperforms many state-of-the-art methods on the single-eye expression recognition task, especially under degraded lighting conditions.

Enhancement of eye socket recognition performance using inverse histogram fusion images and the Gabor transform

AbstractThe eye socket is a cavity in the skull that encloses the eyeball and its surrounding muscles. It has unique shapes in individuals. This study proposes a new recognition method that relies on the eye socket shape and region. This method involves the utilization of an inverse histogram fusion image to generate Gabor features from the identified eye socket regions. These Gabor features are subsequently transformed into Gabor images and employed for recognition by utilizing both traditional methods and deep‐learning models. Four distinct benchmark datasets (Flickr30, BioID, Masked AT & T, and CK+) were used to evaluate the method's performance. These datasets encompass a range of perspectives, including variations in eye shape, covering, and angles. Experimental results and comparative studies indicate that the proposed method achieved a significantly () higher accuracy (average value greater than 92.18%) than that of the relevant identity recognition method and state‐of‐the‐art deep networks (average value less than 78%). We conclude that this improved generalization has significant implications for advancing the methodologies employed for identity recognition.

Intentional looks facilitate faster responding in observers

Humans construct rich representations of other people’s mental states. Here we investigated how intentionality in eye gaze affected perception and responses to gaze. Observers viewed videos of human gazers looking left or right. Unbeknownst to the observers, the gazers could either choose where to look (self-chosen gaze) or were explicitly instructed where to look (computer-instructed gaze). In Experiment 1, observers reported the direction of the gazer’s upcoming look before the eye movement was initiated. Faster responses were found for self-chosen relative to computer-instructed gaze. In Experiments 2 and 3, observers responded by reporting the location of a peripheral target that appeared at the gazed-at or not gazed-at location. Faster responses were found for gazed-at relative to not gazed-at targets and at longer cue-target intervals for self-chosen relative to computer-instructed gaze. The examination of the eye movement kinematics indicated that self-chosen gaze shifts were marked by a larger magnitude of motion within the eye region prior to the eye movement occurring relative to computer-instructed ones. Thus, perceived intentionality in eye gaze facilitates responses in observers with the information about mental states communicated via subtle properties of eye motion.

A Real-Time Embedded System for Driver Drowsiness Detection Based on Visual Analysis of the Eyes and Mouth Using Convolutional Neural Network and Mouth Aspect Ratio.

Currently, the number of vehicles in circulation continues to increase steadily, leading to a parallel increase in vehicular accidents. Among the many causes of these accidents, human factors such as driver drowsiness play a fundamental role. In this context, one solution to address the challenge of drowsiness detection is to anticipate drowsiness by alerting drivers in a timely and effective manner. Thus, this paper presents a Convolutional Neural Network (CNN)-based approach for drowsiness detection by analyzing the eye region and Mouth Aspect Ratio (MAR) for yawning detection. As part of this approach, endpoint delineation is optimized for extraction of the region of interest (ROI) around the eyes. An NVIDIA Jetson Nano-based device and near-infrared (NIR) camera are used for real-time applications. A Driver Drowsiness Artificial Intelligence (DD-AI) architecture is proposed for the eye state detection procedure. In a performance analysis, the results of the proposed approach were compared with architectures based on InceptionV3, VGG16, and ResNet50V2. Night-Time Yawning-Microsleep-Eyeblink-Driver Distraction (NITYMED) was used for training, validation, and testing of the architectures. The proposed DD-AI network achieved an accuracy of 99.88% with the NITYMED test data, proving superior to the other networks. In the hardware implementation, tests were conducted in a real environment, resulting in 96.55% and 14 fps on average for the DD-AI network, thereby confirming its superior performance.

An investigation on the face inversion effect in deaf children.

The face inversion effect is an important indicator of holistic face perception and reflects the developmental level of face processing. This study examined the face inversion effect in deaf or hard of hearing (DHH) children aged 7-17 using the face dimensions task. This task uses photographic images of a face, in which configural and featural information in the eye and mouth regions have been parametrically and independently manipulated. The study aimed to discuss the effect of face inversion on facial processing in DHH children, including two aspects of information processing types (configural versus featural) and processing regions (eyes versus mouth) and compared the results with hearing children. The results revealed that DHH children aged 7-17years exhibit significant face inversion effect, with disruptions observed in both the featural and configural processing of eyes and mouths when faces were inverted. Configural processing was more affected by inversion than featural processing in all children, with larger differences observed in DHH children than in hearing children. This supports the dual-mode hypothesis of holistic face processing. Age correlations were observed in the sensitivity of DHH children to face inversion effect but not among hearing children. The inversion effect of configural mouth processing decreases with age in DHH children.

Tensile properties of glaucomatous human sclera, optic nerve, and optic nerve sheath.

We characterized the tensile behavior of sclera, optic nerve (ON), and ON sheath in eyes from donors with glaucoma, for comparison with published data without glaucoma. Twelve freshly harvested eyes were obtained from donors with history of glaucoma, of average age 86 ± 7 (standard deviation) years. Rectangular samples were taken from anterior, equatorial, posterior, and peripapillary sclera, and ON sheath, while ON was in native form and measured using calipers. Under physiological temperature and humidity, tissues were preconditioned at 5% strain before loading at 0.1mm/s. Force-displacement data were converted into engineering stress-strain curves fit by reduced polynomial hyperelastic models and analyzed by tangent moduli at 3% and 7% strain. Data were compared with an age-matched sample of 7 published control eyes. Optic atrophy was supported by significant reduction in ON cross section to 73% of normal in glaucomatous eyes. Glaucomatous was significantly stiffer than control in equatorial and peripapillary regions (P < 0.001). However, glaucomatous ON and sheath were significantly less stiff than control, particularly at low strain (P < 0.001). Hyperelastic models were well fit to stress-strain data (R2 > 0.997). Tangent moduli had variability similar to control in most regions, but was abnormally large in peripapillary sclera. Tensile properties were varied independently among various regions of the same eyes. Glaucomatous sclera is abnormally stiff, but the ON and sheath are abnormally compliant. These abnormalities correspond to properties predicted by finite element analysis to transfer potentially pathologic stress to the vulnerable disk and lamina cribrosa region during adduction eye movement.

Association of choroidal thickness and blood flow features with asymmetric axial lengths in children with unilateral myopic anisometropia

BackgroundConsidering that changes in the choroidal thickness are closely related to ocular growth, we studied the choroidal thickness (CT) and the blood flow features in children with unilateral myopic anisometropia (UMA) as well as investigating the relationship between choroidal changes and myopia.MethodsSubjective refractive, axial length (AL), and biometric parameters were measured in 98 UMA children (age: 8–15 years). CT and choroidal blood-flow features, including the choroidal vessel volume (CVV), choroidal vascularity index (CVI), and choriocapillaris perfusion area (CCPA), were measured through swept-source optical coherence tomography angiography. The macular region was categorized into four concentric circles of diameters 0–1 mm (central fovea), 1–3 mm (parafovea), 3–6 mm (perifovea), and 6–9 mm (extended), and further categorized into superior (S), inferior (I), temporal (T), and nasal (N) quadrants.ResultsThe aforementioned four regions of myopic eyes displayed significantly lower CT, CVV, and CVI than those of non-myopic eyes. CCPA changes differed across different regions of both the eyes (parts of N and T quadrants). There was an inverse association between CT and the interocular AL difference (central and other regions S, T quadrant). No correlation was noted between CVV and CVI with interocular AL difference. CT and CVV were positively correlated in the 0–6-mm macular region of myopic eyes (Spearman correlation coefficient = 0.763, P < 0.001).ConclusionsIn UMA children, CCT and blood flow may be related to myopia progression. A robust correlation between CT and CVV in the 0–6-mm macular region and reduced CT and diminished blood flow indicated an association with myopia.

Pupil contagion variation with gaze, arousal, and autistic traits

Pupillary contagion occurs when one’s pupil size unconsciously adapts to the pupil size of an observed individual and is presumed to reflect the transfer of arousal. Importantly, when estimating pupil contagion, low level stimuli properties need to be controlled for, to ensure that observations of pupillary changes are due to internal change in arousal rather than the external differences between stimuli. Here, naturalistic images of children’s faces depicting either small or large pupils were presented to a group of children and adolescents with a wide range of autistic traits, a third of whom had been diagnosed with autism. We examined the extent to which pupillary contagion reflects autonomic nervous system reaction through pupil size change, heart rate and skin conductance response. Our second aim was to determine the association between arousal reaction to stimuli and degree of autistic traits. Results show that pupil contagion and concomitant heart rate change, but not skin conductance change, was evident when gaze was restricted to the eye region of face stimuli. A positive association was also observed between pupillary contagion and autistic traits when participants’ gaze was constrained to the eye region. Findings add to a broader understanding of the mechanisms underlying pupillary contagion and its association with autism.

Enhancing Retina Images by Lowpass Filtering Using Binomial Filter.

This study presents a method to enhance the contrast and luminosity of fundus images with boundary reflection. In this work, 100 retina images taken from online databases are utilized to test the performance of the proposed method. First, the red, green and blue channels are read and stored in separate arrays. Then, the area of the eye also called the region of interest (ROI) is located by thresholding. Next, the ratios of R to G and B to G at every pixel in the ROI are calculated and stored along with copies of the R, G and B channels. Then, the RGB channels are subjected to average filtering using a 3 × 3 mask to smoothen the RGB values of pixels, especially along the border of the ROI. In the background brightness estimation stage, the ROI of the three channels is filtered by binomial filters (BFs). This step creates a background brightness (BB) surface of the eye region by levelling the foreground objects like blood vessels, fundi, optic discs and blood spots, thus allowing the estimation of the background illumination. In the next stage, using the BB, the luminosity of the ROI is equalized so that all pixels will have the same background brightness. This is followed by a contrast adjustment of the ROI using CLAHE. Afterward, details of the adjusted green channel are enhanced using information from the adjusted red and blue channels. In the color correction stage, the intensities of pixels in the red and blue channels are adjusted according to their original ratios to the green channel before the three channels are reunited. The resulting color image resembles the original one in color distribution and tone but shows marked improvement in luminosity and contrast. The effectiveness of the approach is tested on the test images and enhancement is noticeable visually and quantitatively in greyscale and color. On average, this method manages to increase the contrast and luminosity of the images. The proposed method was implemented using MATLAB R2021b on an AMD 5900HS processor and the average execution time was less than 10 s. The performance of the filter is compared to those of two other filters and it shows better results. This technique can be a useful tool for ophthalmologists who perform diagnoses on the eyes of diabetic patients.

Joint pyramidal perceptual attention and hierarchical consistency constraint for gaze estimation

Eye gaze provides valuable cues about human intent, making gaze estimation a hot topic. Extracting multi-scale information has recently proven effective for gaze estimation in complex scenarios. However, existing methods for estimating gaze based on multi-scale features tend to focus only on information from single-level feature maps. Furthermore, information across different scales may also lack relevance. To address these issues, we propose a novel joint pyramidal perceptual attention and hierarchical consistency constraint (PaCo) for gaze estimation. The proposed PaCo consists of two main components: pyramidal perceptual attention module (PPAM) and hierarchical consistency constraint (HCC). Specifically, PPAM first extracts multi-scale spatial features using a pyramid structure, and then aggregates information from coarse granularity to fine granularity. In this way, PPAM enables the model to simultaneously focus on both the eye region and facial region at multiple scales. Then, HCC makes constrains consistency on low-level and high-level features, aiming to enhance the gaze semantic consistency between different feature levels. With the combination of PPAM and HCC, PaCo can learn more discriminative features in complex situations. Extensive experimental results show that PaCo achieves significant performance improvements on challenging datasets such as Gaze360, MPIIFaceGaze, and RT-GENE,reducing errors to 10.27°, 3.23°, 6.46°, respectively.

Production of X-ray absorbing ceramics and products based on them for radiotherapy of the periorbital eye region

The paper presents an innovative method for producing high-density ceramics based on bioinert tantalum oxide (Ta2O5) and eye shield products utilizing spark plasma sintering (SPS) technology. Eye shields are in high demand for protecting the cornea and other parts of the eye from ionizing radiation during periorbital eye procedures, such as laser therapy, radiotherapy, and Mohs micrographic surgery (MMS). The dynamics of consolidation of microcrystalline Ta2O5 powder under spark plasma heating conditions in the temperature range of 1000–1400 °C were investigated. The structure, phase, and elemental composition of polycrystalline ceramic samples, including the calculation of crystal structure parameters, were studied using XRD, SEM, and EDS methods. The optimal SPS parameters for achieving maximum relative density (up to 100 %) and Vickers microhardness (up to 510 H V) of the obtained ceramics were determined. The performance properties of the ceramics were examined under conditions of a session of short-focus radiation therapy applied to the periorbital facial area. A dosimetric assessment of the absorbing capacity of the ceramics under X-ray irradiation in modes from 40 to 120 keV was conducted. A prototype of a ceramic eye shield product based on Ta2O5 was developed and manufactured for the first time using SPS technology.

Multistream Gaze Estimation With Anatomical Eye Region Isolation by Synthetic to Real Transfer Learning

Multistream Gaze Estimation With Anatomical Eye Region Isolation by Synthetic to Real Transfer Learning

Severe Circulatory Disturbance in Optic Disc, Retina, and Choroid Following Sub-Tenon Triamcinolone Acetonide Injection for Posterior Scleritis.

To report a rare case of optic disc and chorioretinal vascular occlusions in the right eye following sub-Tenon injection of triamcinolone acetonide (STTA) in an older male patient with scleritis and to discuss the associated risk factors and potential alternative therapies. The medical history and clinical examination, including fundoscopy, fluorescein angiography (FA), and indocyanine green angiography (ICGA), were conducted to diagnose and evaluate the extent of vascular occlusion after STTA. An extensive literature review was undertaken to ascertain the associated risks and consider other treatment options for scleritis. Severe circulatory disturbances in the optic disc and chorioretinal regions of the right eye developed after STTA for exacerbated scleritis, as confirmed by FA and ICGA, resulting in a significant decrease in visual acuity. A complex medical history encompassing ophthalmic and systemic health issues may have contributed to this negative outcome. A literature review pointed out the potential complications of STTA and highlighted alternative approaches to scleritis management. Cautious application of STTA in treating scleritis is crucial, especially in patients with pre-existing vascular or ocular conditions. Thus, a multidisciplinary approach and a careful assessment of various treatment options are vital to minimize the risk of severe complications and improve patient outcomes.

Comparative analysis of macular characteristics in mCNV and contralateral eyes.

To illustrate the characteristics of perforating scleral vessels in macular regions between mCNV eyes and contralateral eyes in unilateral mCNV patients. This was a retrospective study that included patients with unilateral naive mCNV. The study aimed to identify and analyze the distribution of perforating scleral vessels (PSVs) in the macular region of mCNV eyes and contralateral eyes. The central macular choroidal thicknesses (mChT) were measured using optical coherence tomography angiography (OCTA). The grades of myopic atrophic maculopathy (MAM) and macular myopic diffuse chorioretinal atrophy (DCA) were evaluated within groups. The number of PSVs and mChT were compared between contralateral and mCNV eyes based on the grade of DCA. The ROC curves were utilized to explore the diagnostic indexes for mCNV. A total of 102 eyes from 51 patients with unilateral mCNV were included. There was no significance in the severity of MAM or the grade of DCA between mCNV eyes and contralateral eyes (p = 0.074, p = 0.054, respectively). The mean number of PSVs in mCNV eyes was fewer than the contralateral eyes [1.00 (1.00-2.00) vs. 2.00 (0.75-3.00), p = 0.030]. The mChT in mCNV eyes was thinner than the contralateral eyes [36.00 (25.00-53.75) μm vs. 46.00 (31.00-75.25) μm, p = 0.001]. The mean grade of DCA in mCNV eyes was higher than that in contralateral eyes [3.00 (3.00-3.00) vs. 3.00 (2.00-3.00), p = 0.004]. When DCA involved the macular region, there were more PSVs in contralateral eyes than in mCNV eyes [1.50 (1.00-2.00) vs. 2.00 (1.00-3.00), p = 0.042]. Similarly, when DCA involved the foveal region, there were more PSVs in contralateral eyes than in mCNV eyes [1.50 (1.00-2.00) vs. 3.00 (2.00-4.00), p = 0.004]. The grade of DCA and mChT were valuable factors for predicting mCNV eyes (AUC = 0.6566, p = 0.021; AUC = 0.6304, p = 0.029; respectively). When the extent of DCA exceeded the foveal region, the count of PSVs was a good diagnostic factor for predicting mCNV (AUC = 0.7430, p = 0.003). The mean amount of PSVs was significantly lower in the mCNV eyes compared to the contralateral eyes. When the extent of DCA exceeded the foveal region, the count of PSVs was a good diagnostic factor for predicting mCNV. Myopic eyes with a higher grade of DCA and a thinner mChT were more likely to develop mCNV.

Quality non-destructive sorting of large yellow croaker based on image recognition

The YOLOv7 algorithm was used to establish a fast and non-destructive quality identification model for large yellow croaker (Larimichthys crocea) in this study. The cross multi-head attention mechanism in the swin transformer was incorporated into the neck of YOLOv7 architecture to enhance the recognition performance. The freshness classification model based on the total volatile basic nitrogen value was evaluated by freshness indicators, visual features, and texture profile analysis (TPA). The findings indicated that the enhanced model achieved an accuracy rate of 98.6% in freshness classification, which was higher than 85.6% of the original model. Visual features (fish-eye plumpness and turbidity) were highly correlated with all freshness indexes (all above 0.8). The accuracy of freshness discrimination in different lighting environments was also greater than 90%. These results collectively indicate the potential for the eye region images to serve as a reliable indicator for the sorting of freshness in large yellow croakers.