Enhanced prediction of anisotropic deformation behavior using machine learning with data augmentation

Mg alloys possess an inherent plastic anisotropy owing to the selective activation of deformation mechanisms depending on the loading condition.This characteristic results in a diverse range of flow curves that vary with a deformation condition.This study proposes a novel approach for accurately predicting an anisotropic deformation behavior of wrought Mg alloys using machine learning(ML)with data augmentation.The developed model combines four key strategies from data science:learning the entire flow curves,generative adversarial networks(GAN),algorithm-driven hyperparameter tuning,and gated recurrent unit(GRU)architecture.The proposed model,namely GAN-aided GRU,was extensively evaluated for various predictive scenarios,such as interpolation,extrapolation,and a limited dataset size.The model exhibited significant predictability and improved generalizability for estimating the anisotropic compressive behavior of ZK60 Mg alloys under 11 annealing conditions and for three loading directions.The GAN-aided GRU results were superior to those of previous ML models and constitutive equations.The superior performance was attributed to hyperparameter optimization,GAN-based data augmentation,and the inherent predictivity of the GRU for extrapolation.As a first attempt to employ ML techniques other than artificial neural networks,this study proposes a novel perspective on predicting the anisotropic deformation behaviors of wrought Mg alloys.

Defect Detection Model Using Time Series Data Augmentation and Transformation

Time-series data provide important information in many fields,and their processing and analysis have been the focus of much research.However,detecting anomalies is very difficult due to data imbalance,temporal dependence,and noise.Therefore,methodologies for data augmentation and conversion of time series data into images for analysis have been studied.This paper proposes a fault detection model that uses time series data augmentation and transformation to address the problems of data imbalance,temporal dependence,and robustness to noise.The method of data augmentation is set as the addition of noise.It involves adding Gaussian noise,with the noise level set to 0.002,to maximize the generalization performance of the model.In addition,we use the Markov Transition Field(MTF)method to effectively visualize the dynamic transitions of the data while converting the time series data into images.It enables the identification of patterns in time series data and assists in capturing the sequential dependencies of the data.For anomaly detection,the PatchCore model is applied to show excellent performance,and the detected anomaly areas are represented as heat maps.It allows for the detection of anomalies,and by applying an anomaly map to the original image,it is possible to capture the areas where anomalies occur.The performance evaluation shows that both F1-score and Accuracy are high when time series data is converted to images.Additionally,when processed as images rather than as time series data,there was a significant reduction in both the size of the data and the training time.The proposed method can provide an important springboard for research in the field of anomaly detection using time series data.Besides,it helps solve problems such as analyzing complex patterns in data lightweight.

Perpendicular-Cutdepth:Perpendicular Direction Depth Cutting Data Augmentation Method

Depth estimation is an important task in computer vision.Collecting data at scale for monocular depth estimation is challenging,as this task requires simultaneously capturing RGB images and depth information.Therefore,data augmentation is crucial for this task.Existing data augmentationmethods often employ pixel-wise transformations,whichmay inadvertently disrupt edge features.In this paper,we propose a data augmentationmethod formonocular depth estimation,which we refer to as the Perpendicular-Cutdepth method.This method involves cutting realworld depth maps along perpendicular directions and pasting them onto input images,thereby diversifying the data without compromising edge features.To validate the effectiveness of the algorithm,we compared it with existing convolutional neural network(CNN)against the current mainstream data augmentation algorithms.Additionally,to verify the algorithm’s applicability to Transformer networks,we designed an encoder-decoder network structure based on Transformer to assess the generalization of our proposed algorithm.Experimental results demonstrate that,in the field of monocular depth estimation,our proposed method,Perpendicular-Cutdepth,outperforms traditional data augmentationmethods.On the indoor dataset NYU,our method increases accuracy from0.900 to 0.907 and reduces the error rate from0.357 to 0.351.On the outdoor dataset KITTI,our method improves accuracy from 0.9638 to 0.9642 and decreases the error rate from 0.060 to 0.0598.

Image segmentation of exfoliated two-dimensional materials by generative adversarial network-based data augmentation

Mechanically cleaved two-dimensional materials are random in size and thickness.Recognizing atomically thin flakes by human experts is inefficient and unsuitable for scalable production.Deep learning algorithms have been adopted as an alternative,nevertheless a major challenge is a lack of sufficient actual training images.Here we report the generation of synthetic two-dimensional materials images using StyleGAN3 to complement the dataset.DeepLabv3Plus network is trained with the synthetic images which reduces overfitting and improves recognition accuracy to over 90%.A semi-supervisory technique for labeling images is introduced to reduce manual efforts.The sharper edges recognized by this method facilitate material stacking with precise edge alignment,which benefits exploring novel properties of layered-material devices that crucially depend on the interlayer twist-angle.This feasible and efficient method allows for the rapid and high-quality manufacturing of atomically thin materials and devices.

YOLO-Based Damage Detection with StyleGAN3 Data Augmentation for Parcel Information-Recognition System

Damage to parcels reduces customer satisfactionwith delivery services and increases return-logistics costs.This can be prevented by detecting and addressing the damage before the parcels reach the customer.Consequently,various studies have been conducted on deep learning techniques related to the detection of parcel damage.This study proposes a deep learning-based damage detectionmethod for various types of parcels.Themethod is intended to be part of a parcel information-recognition systemthat identifies the volume and shipping information of parcels,and determines whether they are damaged;this method is intended for use in the actual parcel-transportation process.For this purpose,1)the study acquired image data in an environment simulating the actual parcel-transportation process,and 2)the training dataset was expanded based on StyleGAN3 with adaptive discriminator augmentation.Additionally,3)a preliminary distinction was made between the appearance of parcels and their damage status to enhance the performance of the parcel damage detection model and analyze the causes of parcel damage.Finally,using the dataset constructed based on the proposed method,a damage type detection model was trained,and its mean average precision was confirmed.This model can improve customer satisfaction and reduce return costs for parcel delivery companies.

Change in self-image pressure level before and after autologous fat breast augmentation and its effect on social adaptability

BACKGROUND There is an increasingly strong demand for appearance and physical beauty in social life,marriage,and other aspects with the development of society and the improvement of material living standards.An increasing number of people have improved their appearance and physical shape through aesthetic plastic surgery.The female breast plays a significant role in physical beauty,and droopy or atrophied breasts can frequently lead to psychological inferiority and lack of confidence in women.This,in turn,can affect their mental health and quality of life.AIM To analyze preoperative and postoperative self-image pressure-level changes of autologous fat breast augmentation patients and their impact on social adaptability.METHODS We selected 160 patients who underwent autologous fat breast augmentation at the First Affiliated Hospital of Xinxiang Medical University from January 2020 to December 2022 using random sampling method.The general information,selfimage pressure level,and social adaptability of the patients were investigated using a basic information survey,body image self-assessment scale,and social adaptability scale.The self-image pressure-level changes and their effects on the social adaptability of patients before and after autologous fat breast augmentation were analyzed.RESULTS We collected 142 valid questionnaires.The single-factor analysis results showed no statistically significant difference in the self-image pressure level and social adaptability score of patients with different ages,marital status,and monthly income.However,there were significant differences in social adaptability among patients with different education levels and employment statuses.The correlation analysis results revealed a significant correlation between the self-image pressure level and social adaptability score before and after surgery.Multiple factors analysis results showed that the degree of concern caused by appearance in selfimage pressure,the degree of possible behavioral intervention,the related distress caused by body image,and the influence of body image on social life influenced the social adaptability of autologous fat breast augmentation patients.CONCLUSION The self-image pressure on autologous fat breast augmentation patients is inversely proportional to their social adaptability.

Transcranial direct current stimulation as early augmentation in adolescent obsessive compulsive disorder:A pilot proof-of-concept randomized control trial

BACKGROUND Transcranial direct current stimulation(tDCS)is proven to be safe in treating various neurological conditions in children and adolescents.It is also an effective method in the treatment of OCD in adults.AIM To assess the safety and efficacy of tDCS as an add-on therapy in drug-naive adolescents with OCD.METHODS We studied drug-naïve adolescents with OCD,using a Children’s Yale-Brown obsessive-compulsive scale(CY-BOCS)scale to assess their condition.Both active and sham groups were given fluoxetine,and we applied cathode and anode over the supplementary motor area and deltoid for 20 min in 10 sessions.Reassessment occurred at 2,6,and 12 wk using CY-BOCS.RESULTS Eighteen adolescents completed the study(10-active,8-sham group).CY-BOCS scores from baseline to 12 wk reduced significantly in both groups but change at baseline to 2 wk was significant in the active group only.The mean change at 2 wk was more in the active group(11.8±7.77 vs 5.25±2.22,P=0.056).Adverse effects between the groups were comparable.CONCLUSION tDCS is safe and well tolerated for the treatment of OCD in adolescents.However,there is a need for further studies with a larger sample population to confirm the effectiveness of tDCS as early augmentation in OCD in this population.

Data augmentation of ultrasound imaging for non-invasive white blood cell in vitro peritoneal dialysis

The limited amount of data in the healthcare domain and the necessity of training samples for increased performance of deep learning models is a recurrent challenge,especially in medical imaging.Newborn Solutions aims to enhance its non-invasive white blood cell counting device,Neosonics,by creating synthetic in vitro ultrasound images to facilitate a more efficient image generation process.This study addresses the data scarcity issue by designing and evaluating a continuous scalar conditional Generative Adversarial Network(GAN)to augment in vitro peritoneal dialysis ultrasound images,increasing both the volume and variability of training samples.The developed GAN architecture incorporates novel design features:varying kernel sizes in the generator’s transposed convolutional layers and a latent intermediate space,projecting noise and condition values for enhanced image resolution and specificity.The experimental results show that the GAN successfully generated diverse images of high visual quality,closely resembling real ultrasound samples.While visual results were promising,the use of GAN-based data augmentation did not consistently improve the performance of an image regressor in distinguishing features specific to varied white blood cell concentrations.Ultimately,while this continuous scalar conditional GAN model made strides in generating realistic images,further work is needed to achieve consistent gains in regression tasks,aiming for robust model generalization.

Proactive traffic responsive control based on state-space neural network and extended Kalman filter

The state-space neural network and extended Kalman filter model is used to directly predict the optimal timing plan that corresponds to futuristic traffic conditions in real time with the purposes of avoiding the lagging of the signal timing plans to traffic conditions. Utilizing the traffic conditions in current and former intervals, the network topology of the state-space neural network （SSNN）, which is derived from the geometry of urban arterial routes, is used to predict the optimal timing plan corresponding to the traffic conditions in the next time interval. In order to improve the effectiveness of the SSNN, the extended Kalman filter （EKF） is proposed to train the SSNN instead of conventional approaches. Raw traffic data of the Guangzhou Road, Nanjing and the optimal signal timing plan generated by a multi-objective optimization genetic algorithm are applied to test the performance of the proposed model. The results indicate that compared with the SSNN and the BP neural network, the proposed model can closely match the optimal timing plans in futuristic states with higher efficiency.

The Empirical Analysis on the Dynamic Effect of Rural-urban Migration on the Consumption Growth of Residents in China——Based on Varying Parameter State-space Model

The research constructed varying parameter state-space model and per- formed estimation on dynamic relationship between urban-rural migration and aggre- gate consumption expenditure on basis of dual economic structure. The results showed that urban consumption growth made the most contribution to aggregate consumption growth, followed by urban-rural migration caused consumption. The role of rural consumption growth kept stable, but consumption caused by population growth was decreasing. Therefore, China consumption growth mainly relies on urban consumption expenditure and urban-rural migration.

Evaluation of anatomical considerations in the posterior maxillae for sinus augmentation

The edentulous posterior maxilla is considered a clinical challenge during dental implant treatment for many dental practitioners. This is because its insufficient bone quality, deficient alveolar ridge, spiny ridges, undercuts, and sinus pneumatization are often encountered after tooth loss. To overcome these problems, several approaches have been developed and are currently used, including sinus augmentation and bone augmentation. Today, two main procedures of sinus floor elevation for dental implant placement are in use: a two-stage technique using the lateral window approach, and a onestage technique using a lateral or a crestal approach. In this study, we deal with the anatomic relations ofthe structures of the maxillary sinus during sinus augmentation. These anatomical findings can help in complications and potential injuries of the maxillary sinus procedures. It can be suggested that pre-operative evaluation is helpful for diagnosis and treatment planning and minimizing complication during the surgery.

Complex knee injuries treated in acute phase: Long-term results using Ligament Augmentation and Reconstruction System artificial ligament

AIM To present the long-term results of complex knee injuries, treated early using the Ligament Augmentation and Reconstruction System(LARS) artificial ligament to reconstruct posterior cruciate ligament(PCL).METHODS From September 1997 to June 2010, thirty-eight complex knee injuries were treated, where early arthroscopic PCL reconstructions were undergone, using the LARS(Surgical Implants and Devices, Arc-sur-Tille, France) artificial ligament. Exclusion criteria were: Late(> 4 wk) reconstruction, open technique, isolated PCL reconstruction, knee degenerative disease, combinedfracture or vascular injury and use of allograft or autograft for PCL reconstruction. Clinical and functional outcomes were assessed with IKDC Subjective Knee Form, KOS-ADLS questionnaire, Lysholm scale and SF-12 Health Survey. Posterior displacement(PD) was measured with the Telos Stress Device. RESULTS Seven patients were excluded; two because of coexisting knee osteoarthritis and the remaining five because of failure to attend the final follow-up. The sample consisted of 31 patients with mean age at the time of reconstruction 33.2 ± 12.5 years(range 17-61). The postoperative follow-up was on average 9.27 ± 4.27 years(range 5-18). The mean average IKDC and KOS scores were 79.32 ± 17.1 and 88.1 ± 12.47% respectively. Average PD was 3.61 ± 2.15 mm compared to 0.91 ± 1.17 mm in the uninjured knees(one with grade 1+ and two with grade 2 +). Dial test was found positive in one patient, whereas the quadriceps active drawer test was positive in three patients. None was tested positive on the reverse-pivot shift test. The range of motion(ROM) was normal in thirty knees, in comparison with the contralateral one. There was no extension deficit. Osteoarthritic changes were found in three knees(9.6%).CONCLUSION Early treatment of complex knee injuries, using LARS artificial ligament for PCL reconstruction sufficiently reduces posterior tibia displacement and provides satisfactory long-term functional outcomes.

Sequential Fluorescent Labeling Observation of Maxillary Sinus Augmentation by a Tissue-engineered Bone Complex in Canine Model

Aim To evaluate the effects of maxillary sinus floor elevation by a tissue-engineered bone complex of β-tricalcium phosphate （β-TCP） and autologous osteoblasts in dogs. Methodology Autologous osteoblasts from adult Beagle dogs were cultured in vitro. They were further combined with β-TCP to construct the tissue-engineered bone complex. 12 cases of maxillary sinus floor elevation surgery were made bilaterally in 6 animals and randomly repaired with the following 3 groups of materials： Group A （osteoblasts/D-TCP）; Group B （β-TCP）; Group C （autogenous bone） （n=4 per group）. A polychrome sequential fluorescent labeling was performed post-operatively and the animals were sacrificed 24 weeks after operation for histological observation.Results Our results showed that autologous osteoblasts were successfully expanded and the osteoblastic phenol- types were confirmed by ALP and Alizarin red staining. The cells could attach and proliferate well on the surface of the ~3-TCP scaffold. The fluorescent and histological observation showed that the tissue-engineered bone complex had an earlier mineralization and more bone formation inside the scaffold than β-TCP along or even autologous bone. It had also maximally maintained the elevated sinus height than both control groups. Conclusion Porous 13-TCP has served as a good scaffold for autologous osteoblasts seeding. The tissue-engineered bone complex with β-TCP and autologous osteoblasts might be a better alternative to autologous bone for the clinical edentulous maxillary sinus augmentation.

Computation of satellite clock-ephemeris augmentation parameters for dual-frequency multi-constellation satellite-based augmentation system

Dual-frequency multi-constellation(DFMC) satellitebased augmentation system(SBAS) does not broadcast fast correction, which is important in reducing range error in L1-only SBAS.Meanwhile, the integrity bound of a satellite at low elevation is so loose that the service availability is decreased near the boundary of the service area. Therefore, the computation of satellite clockephemeris(SCE) augmentation parameters needs improvement.We propose a method introducing SCE prediction to eliminate most of the SCE error resulting from global navigation satellite system GNSS broadcast message. Compared with the signal-inspace(SIS) after applying augmentation parameters broadcast by the wide area augmentation system(WAAS), SIS accuracy after applying augmentation parameters computed by the proposed algorithm is improved and SIS integrity is ensured. With global positioning system(GPS) only, the availability of category-I(CAT-I)with a vertical alert level of 15 m in continental United States is about 90%, while the availability in the other part of the WAAS service area is markedly improved. With measurements made by the stations from the crustal movement observation network of China,users in some part of China can obtain CAT-I(vertical alert limit is 15 m) service with GPS and global navigation satellite system(GLONASS).

Repeated lumps and infections: A case report on breast augmentation complications

BACKGROUND Polyacrylamide hydrogel(PAAG)injections were once common in breast augmentation and have been prohibited for augmentation mammaplasty in China since a large number of patients who underwent breast augmentation with PAAG injections have continued to seek medical advice as a result of related complications.Among all these complications,distant migration is relatively rare.CASE SUMMARY A 49-year-old female presented at the hospital with a one-year history of a vulvar lump.The sonography of the lump showed several subcutaneous fluid-filled regions from the left vulva to the pubic symphysis,which suggested possible fat liquefaction.An enhanced magnetic resonance imaging(MRI)revealed a cystic area,which was considered a benign lesion.Intraoperative observations showed that the mass did not have an obvious capsule,the subcutaneous tissue presented as a cavity,and some yellow material came out of this cavity.A culture of the drainage did not show bacterial contamination.Histopathology revealed a foreign body granuloma.After resection and closed drainage,lumps were successively observed in the left lower abdomen and the bilateral hypochondriac region with infections.Sonography found that the hypoechoic areas in the bilateral hypochondriac region seemed continuous with deep in the breasts.The patient reported that she had undergone surgery with PAAG injections 20 years ago after she was repeatedly asked about her past history.Finally,a diagnosis of distant migration of PAAG was made.CONCLUSION PAAG gel can migrate after long periods of time.A diagnosis should not be limited to the area where the symptom develops.

Effects of augmentation of liver regeneration recombinant plasmid on rat hepatic fibrosis

AIM: To investigate the effects of eukaryotic expression of plasmid on augmentation of liver regeneration (ALR) in rat hepatic fibrosis and to explore their mechanisms. METHODS: Ten rats were randomly selected from 50 Wistar rats as normal control group. The rest were administered intraperitoneally with porcine serum twice weekly. After 8 wk, they were randomly divided into: model control group, colchicine group (Col), first ALR group (ALR1), second ALR group (ALR2). Then colchicine ALR recombinant plasmid were used to treat them respectively. At the end of the 4th wk, rats were killed. Serum indicators were detected and histopathological changes were graded. Expression of type Ⅰ, Ⅲ, collagen and TIMP-1 were detected by immunohisto-chemistry and expression of TIMP-1 mRNA was detected by semi-quantified RT-PCR. RESULTS: The histologic examination showed that the degree of the rat hepatic fibrosis in two ALR groups was lower than those in model control group. Compared with model group, ALR significantly reduced the serum levels of ALT, AST, HA, LN, PCIII and IV (P<0.05). Immunohistochemical staining showed that expression of type Ⅰ, Ⅲ, collagen and TIMP-1 in two ALR groups was ameliorated dramatically compared with model group (I collagen: 6.94±1.42,5.80±1.66 and 10.83±3.58 in ALR1, ALR2 and model groups, respectively; Ⅲ collagen: 7.18±1.95, 4.50±1.67 and 10.25±2.61, respectively; TIMP-1: 0.39±0.05,0.20±0.06 and 0.53±0.12, respectively,P<0.05 or P<0.01). The expression level of TIMP-1 mRNA in the liver tissues was markedly decreased in two ALR groups compared with model group (TIMP-1 mRNA/β-actin: 0.89±0.08, 0.65±0.11 and 1.36±0.11 in ALR1, ALR2 and model groups respectively, P<0.01). CONCLUSION: ALR recombinant plasmid has beneficial effects on rat hepatic fibrosis by enhancing regeneration of injured liver cells and inhibiting TIMP-1 expressions.

Amisulpride augmentation therapy improves cognitive performance and psychopathology in clozapine‑resistant treatment‑refractory schizophrenia:a 12‑week randomized,double‑blind,placebo‑controlled trial

Background:Although clozapine is an effective option for treatment-resistant schizophrenia(TRS),there are still 1/3 to 1/2 of TRS patients who do not respond to clozapine.The main purpose of this randomized,double-blind,placebocontrolled trial was to explore the amisulpride augmentation efficacy on the psychopathological symptoms and cognitive function of clozapine-resistant treatment-refractory schizophrenia(CTRS)patients.Methods:A total of 80 patients were recruited and randomly assigned to receive initial clozapine plus amisulpride(amisulpride group)or clozapine plus placebo(placebo group).Positive and Negative Syndrome Scale(PANSS),Scale for the Assessment of Negative Symptoms(SANS),Clinical Global Impression(CGI)scale scores,Repeatable Battery for the Assessment of Neuropsychological Status(RBANS),Treatment Emergent Symptom Scale(TESS),laboratory measurements,and electrocardiograms(ECG)were performed at baseline,week 6,and week 12.Results:Compared with the placebo group,amisulpride group had a lower PANSS total score,positive subscore,and general psychopathology subscore at week 6 and week 12(PBonferroni<0.01).Furthermore,compared with the placebo group,the amisulpride group showed an improved RBANS language score at week 12(PBonferroni<0.001).Amisulpride group had a higher treatment response rate(P=0.04),lower scores of CGI severity and CGI efficacy at week 6 and week 12 than placebo group(PBonferroni<0.05).There were no differences between the groups in body mass index(BMI),corrected QT(QTc)intervals,and laboratory measurements.This study demonstrates that amisulpride augmentation therapy can safely improve the psychiatric symptoms and cognitive performance of CTRS patients.

Recursive State-space Model Identification of Non-uniformly Sampled Systems Using Singular Value Decomposition

In this paper a recursive state-space model identification method is proposed for non-uniformly sampled systems in industrial applications. Two cases for measuring all states and only output(s) of such a system are considered for identification. In the case of state measurement, an identification algorithm based on the singular value decomposition(SVD) is developed to estimate the model parameter matrices by using the least-squares fitting. In the case of output measurement only, another identification algorithm is given by combining the SVD approach with a hierarchical identification strategy. An example is used to demonstrate the effectiveness of the proposed identification method.

Data Augmentation Technology Driven By Image Style Transfer in Self-Driving Car Based on End-to-End Learning

With the advent of deep learning,self-driving schemes based on deep learning are becoming more and more popular.Robust perception-action models should learn from data with different scenarios and real behaviors,while current end-to-end model learning is generally limited to training of massive data,innovation of deep network architecture,and learning in-situ model in a simulation environment.Therefore,we introduce a new image style transfer method into data augmentation,and improve the diversity of limited data by changing the texture,contrast ratio and color of the image,and then it is extended to the scenarios that the model has been unobserved before.Inspired by rapid style transfer and artistic style neural algorithms,we propose an arbitrary style generation network architecture,including style transfer network,style learning network,style loss network and multivariate Gaussian distribution function.The style embedding vector is randomly sampled from the multivariate Gaussian distribution and linearly interpolated with the embedded vector predicted by the input image on the style learning network,which provides a set of normalization constants for the style transfer network,and finally realizes the diversity of the image style.In order to verify the effectiveness of the method,image classification and simulation experiments were performed separately.Finally,we built a small-sized smart car experiment platform,and apply the data augmentation technology based on image style transfer drive to the experiment of automatic driving for the first time.The experimental results show that:(1)The proposed scheme can improve the prediction accuracy of the end-to-end model and reduce the model’s error accumulation;(2)the method based on image style transfer provides a new scheme for data augmentation technology,and also provides a solution for the high cost that many deep models rely heavily on a large number of label data.

Image Augmentation-Based Food Recognition with Convolutional Neural Networks

Image retrieval for food ingredients is important work,tremendously tiring,uninteresting,and expensive.Computer vision systems have extraordinary advancements in image retrieval with CNNs skills.But it is not feasible for small-size food datasets using convolutional neural networks directly.In this study,a novel image retrieval approach is presented for small and medium-scale food datasets,which both augments images utilizing image transformation techniques to enlarge the size of datasets,and promotes the average accuracy of food recognition with state-of-the-art deep learning technologies.First,typical image transformation techniques are used to augment food images.Then transfer learning technology based on deep learning is applied to extract image features.Finally,a food recognition algorithm is leveraged on extracted deepfeature vectors.The presented image-retrieval architecture is analyzed based on a smallscale food dataset which is composed of forty-one categories of food ingredients and one hundred pictures for each category.Extensive experimental results demonstrate the advantages of image-augmentation architecture for small and medium datasets using deep learning.The novel approach combines image augmentation,ResNet feature vectors,and SMO classification,and shows its superiority for food detection of small/medium-scale datasets with comprehensive experiments.