In era of big data, the computer vision-assisted textual extraction techniques for financial invoices have been a major concern. Currently, such tasks are mainly implemented via traditional image processing techniques. However, they highly rely on manual feature extraction and are mainly developed for specific financial invoice scenes. The general applicability and robustness are the major challenges faced by them. As consequence, deep learning can adaptively learn feature representation for different scenes and be utilized to deal with the above issue. As a consequence, this work introduces a classic pre-training model named visual transformer to construct a lightweight recognition model for this purpose. First, we use image processing technology to preprocess the bill image. Then, we use a sequence transduction model to extract information. The sequence transduction model uses a visual transformer structure. In the stage target location, the horizontal-vertical projection method is used to segment the individual characters, and the template matching is used to normalize the characters. In the stage of feature extraction, the transformer structure is adopted to capture relationship among fine-grained features through multi-head attention mechanism. On this basis, a text classification procedure is designed to output detection results. Finally, experiments on a real-world dataset are carried out to evaluate performance of the proposal and the obtained results well show the superiority of it. Experimental results show that this method has high accuracy and robustness in extracting financial bill information.
Citation: Tao Wang, Min Qiu. A visual transformer-based smart textual extraction method for financial invoices[J]. Mathematical Biosciences and Engineering, 2023, 20(10): 18630-18649. doi: 10.3934/mbe.2023826
| [1] | Nur Jassriatul Aida binti Jamaludin, Shanmugan Subramani . Thermal performance of LED fixed on CVD processed ZnO thin film on Al substrates at various O2 gas flow rates. AIMS Materials Science, 2018, 5(2): 246-256. doi: 10.3934/matersci.2018.2.246 |
| [2] | Nur Jassriatul Aida binti Jamaludin, Shanmugan Subramani, Mutharasu Devarajan . Thermal and optical performance of chemical vapor deposited zinc oxide thin film as thermal interface material for high power LED. AIMS Materials Science, 2018, 5(3): 402-413. doi: 10.3934/matersci.2018.3.402 |
| [3] | Muhamed Shajudheen V P, Saravana Kumar S, Senthil Kumar V, Uma Maheswari A, Sivakumar M, Sreedevi R Mohan . Enhancement of anticorrosion properties of stainless steel 304L using nanostructured ZnO thin films. AIMS Materials Science, 2018, 5(5): 932-944. doi: 10.3934/matersci.2018.5.932 |
| [4] | Haridas D. Dhaygude, Surendra K. Shinde, Ninad B. Velhal, G.M. Lohar, Vijay J. Fulari . Synthesis and characterization of ZnO thin film by low cost modified SILAR technique. AIMS Materials Science, 2016, 3(2): 349-356. doi: 10.3934/matersci.2016.2.349 |
| [5] | Haiqing Yao, Fei Li, Jodie Lutkenhaus, Masaya Kotaki, Hung-Jue Sue . High-performance photocatalyst based on nanosized ZnO-reduced graphene oxide hybrid for removal of Rhodamine B under visible light irradiation. AIMS Materials Science, 2016, 3(4): 1410-1425. doi: 10.3934/matersci.2016.4.1410 |
| [6] | Chaitra Srikanth, Gattumane Motappa Madhu, Hemanth Bhamidipati, Siddarth Srinivas . The effect of CdO–ZnO nanoparticles addition on structural, electrical and mechanical properties of PVA films. AIMS Materials Science, 2019, 6(6): 1107-1123. doi: 10.3934/matersci.2019.6.1107 |
| [7] | Sironmani Palraj, Muthaiah Selvaraj, Sundarajan Muthukrishnan . Effect of pretreatments on electrodeposited epoxy coatings for electronic industries. AIMS Materials Science, 2016, 3(1): 214-230. doi: 10.3934/matersci.2016.1.214 |
| [8] | Iryna Markevich, Tetyana Stara, Larysa Khomenkova, Volodymyr Kushnirenko, Lyudmyla Borkovska . Photoluminescence engineering in polycrystalline ZnO and ZnO-based compounds. AIMS Materials Science, 2016, 3(2): 508-524. doi: 10.3934/matersci.2016.2.508 |
| [9] | Jens Bürger, Alireza Goudarzi, Darko Stefanovic, Christof Teuscher . Computational capacity and energy consumption of complex resistive switch networks. AIMS Materials Science, 2015, 2(4): 530-545. doi: 10.3934/matersci.2015.4.530 |
| [10] | Clotaire Chevalier-César, Martine Capochichi-Gnambodoe, Fang Lin, Dapeng Yu, Yamin Leprince-Wang . Effect of growth time and annealing on the structural defect concentration of hydrothermally grown ZnO nanowires. AIMS Materials Science, 2016, 3(2): 562-572. doi: 10.3934/matersci.2016.2.562 |
In era of big data, the computer vision-assisted textual extraction techniques for financial invoices have been a major concern. Currently, such tasks are mainly implemented via traditional image processing techniques. However, they highly rely on manual feature extraction and are mainly developed for specific financial invoice scenes. The general applicability and robustness are the major challenges faced by them. As consequence, deep learning can adaptively learn feature representation for different scenes and be utilized to deal with the above issue. As a consequence, this work introduces a classic pre-training model named visual transformer to construct a lightweight recognition model for this purpose. First, we use image processing technology to preprocess the bill image. Then, we use a sequence transduction model to extract information. The sequence transduction model uses a visual transformer structure. In the stage target location, the horizontal-vertical projection method is used to segment the individual characters, and the template matching is used to normalize the characters. In the stage of feature extraction, the transformer structure is adopted to capture relationship among fine-grained features through multi-head attention mechanism. On this basis, a text classification procedure is designed to output detection results. Finally, experiments on a real-world dataset are carried out to evaluate performance of the proposal and the obtained results well show the superiority of it. Experimental results show that this method has high accuracy and robustness in extracting financial bill information.
Resistive random access memory (ReRAM) has been regarded as next generation nonvolatile memory due to its simple structure and compatibility with complementary metal oxide semiconductor technology [1,2,3,4,5]. Contrary to traditional flash memories, ReRAM offer some unique advantages of faster writing rate, smaller bit size, and lower operating voltages. More importantly, by imposing electric pulses the transition in resistive states can be achieved within tenths of nanoseconds at room temperature and the resultant resistance states can be retained for 10 years [2,6,7].
A typical configuration of the ReRAM device usually consists of metal/insulator/metal sandwich like structure and its operation is based on switching of resistance states from high to low or vice versa by putting voltage stimuli. To date, variety of materials has been exploited for resistive switching characteristics. Among them all, binary transition metal oxide (TiO2, NiO, ZnO, and Cu2O) [1,2,4,6,8,9,10,11] attracted increasing interests in material choice due to controllable compositions. Generally, the switching behaviours can be classified into (1) unipolar and (2) bipolar switching. The unipolar resistive switching (RS) is independent of bias polarity and strongly dependent on strength of input voltage signal. While, bipolar switching shows a directional RS depending on the polarity of the applied voltage [12].
Zinc oxide (ZnO) has been widely well known material for its excellent optical, electronics and sensing applications. In recent past, it has also been exploited as resistive switching material too [13,14]. In this manuscript, we deployed a state of the art electrochemical deposition route to synthesize ZnO thin films. Electrodeposition is an effective way to fabricate high quality metal oxide thin films, at low temperature and ambient atmosphere. Moreover, in this process, the deposition of metal oxide layers on the substrate is driven by the external electric field. Therefore, it is facile to precisely control the layer microstructures by this method and further to design heterostructures with novel functionalities. Herein, the effect of deposition temperature on crystal structure, surface morphology and specifically resistive switching characteristics was investigated. It was found that crystallinity and surface morphologies can be well tuned by adjusting the deposition temperature. Furthermore, the samples prepared at higher temperatures demonstrated excellent resistive switching performances. To explain RS behaviour in these films, a conduction model based on conducting filaments was proposed whose formation/rupture is attributed to change the device resistive states.
Electrodeposition of ZnO was carried out using an Autolab 302N electrochemical workstation. A standard three-electrode setup in an undivided cell was used. Fluorine doped Tin oxide (FTO) (9.3-9.7 Ω, Asashi Glass Corporation, Japan, 1.1 mm × 26 mm × 30 mm) was used as the working electrode, while platinum foil (0.2 mm × 10 mm × 20 mm) was used as the counter electrode. The distance between the two electrodes was 30 mm. The reference electrode was an Ag/AgCl electrode in 4 M KCl solution, against which all the potentials reported herein were measured.
The FTO substrates were first cleaned by ethanol and acetone followed by treated by sonication/ultra-violet ozone. The electrodeposition was carried out in a solution of 0.01 M Zn(NO3)2∙6H2O at 1 mA for 20 min, at 30, 50 and 75 °C. The phase composition of the samples was characterized by X-ray diffraction (Philips X’pert Multipurpose X-ray Diffraction System (MRD) with Cu Kα). The surface morphologies of the samples were observed by scanning electron microscopy (SEM, Nova 230). To measure the electrical properties of the prepared nanocomposite films, Au top electrodes (250 μm diameter) were patterned and deposited by sputtering using a metal shadow mask. Voltage-current curves of the films were measured using an Autolab 302N electrochemical workstation controlled with Nova software. During measurement, the working electrode and sensor electrode were connected to the top Au electrode, while the reference and counter electrode were connected to the FTO substrate. A schematic of the I-V measurement setup is shown in Figure 1.
Figure 1. Setup for the current-voltage measurements of ZnO thin films using an electrochemical workstation.
The electrochemical synthesis of ZnO is a four step process: First, nitrate ions and H2O are electrochemically reduced at the surface of the working electrode, resulting in an increase in local pH value in the vicinity of the reference electrode (equation (1), (2)); Then, the increase in the local pH leads to the precipitation of zinc ions as zinc hydroxide (Zn(OH)2, equation (3)) at suitable temperature and Zn(OH)2 can be transformed into ZnO.
| NO3- + H2O + 2e- → NO2- + 2OH- | (1)—0.20 V vs. Ag/AgCl |
| 2H2O + 2e- → H2 + 2OH- | (2)—1.05 V vs. Ag/AgCl |
| Zn2+ + 2OH- → Zn(OH)2 | (3) |
| Zn(OH)2 → ZnO + H2O | (4) |
The X-ray diffraction pattern of ZnO thin films synthesized on FTO substrate at 30 and 75 °C is shown in figure 2. All diffraction peaks are consistent with the FTO and ZnO wurtzite structure, which can be indexed to a standard spectrum of JCPDS (No. 36-1451). In the XRD patterns, the (002) peak of sample prepared at 75 °C is dominant, and its intensity is much higher than that of the other peaks and of sample prepared at 20 °C. This also points towards its preferential growth in [001] orientation along c-axis and improved crystallinity over the sample prepared at low temperatures.
Figure 2. XRD patterns of ZnO thin films at 30 °C and 75 °C on FTO substrate.The SEM images of ZnO thin films prepared at 30, 50 and 75 °C can be depicted in Figure 3. It can be clearly observed that by increasing deposition temperature, the surface morphology of ZnO films was transformed from nano flakes (figure 3a) to agglomerated nanoparticles (figure 3b) and then to nanorods arrays (figure 3c). These nanorods are well-crystalline and their average diameter is measured as about 250 nm. Usually, ZnO grows preferentially along [001] direction in aqueous solution due to lowest surface energy of (001) facet. And the growth velocity along [010] directions is slower than that of [001] direction, resulting in nanorod morphology.
Figure 3. Scanning electron microscopy of ZnO thin films synthesized at (a) 30 °C (b) 50 °C and (c) 75 °C.The typical current-voltage curves of ZnO thin films synthesized at various temperatures are shown in figure 4. To carrying out such measurements, the dc voltage sweep mode was utilized in the sequence of 0 à −3 à 0 à +3à 0 V, at a scan rate of 0.01 V/s. During the measurements, the bias voltage was applied on the top electrode (TE) with bottom electrode (BE) as grounded, and neither a forming process nor a current compliance was necessary for activating the memory effort. It is interesting to note that samples prepared at low temperatures (30 and 50 °C) didn’t show any considerable resistive switching phenomenon as shown in figure 4. For the ZnO thin films prepared at 75 °C (ZnO nanorods), a jump in resistance state from high resistance to low resistance state at about −2.2V which is referred as the set process. Subsequently, a reset process (~2.5 V) was also observed by sweeping the voltage in positive direction. It is noteworthy to mention that our devices (ZnO nanorods) exhibit typical bipolar resistive switching, that is, the cell can only be written with a negative bias and erased with a positive one which was verified by numerous measurements.
Figure 4. Typical current-voltage curves of ZnO thin films prepared at various temperatures in semi-logarithmic scale.
The endurance characteristics of the Au/ZnO(nanorods)/FTO memory cell are shown in figure 5. The memory window defined by the two resistance states, i.e., (ROFF − RON)/RON ≈ ROFF/RON. The device successfully demonstrated to distinguish between two resistive states for about 2000 cycles. This memory margin usually utilized to distinguish the storage information between “1” and “0”. The resistance of the HRS scatter in a certain extent during cycling. However, due to long cycle numbers, this kind of scattering may be tolerated. The endurance measurements ensured that the switching between ON and OFF states is highly controllable, reversible, and reproducible. After the device was switched ON or OFF, no electrical power was needed to maintain the resistance within the given state.
Figure 5. Data endurance performance of ZnO nanorods based memory device for 2000 cycles.
Subsequently, the data retention test was also carried out in both the OFF and ON states for 1800 seconds at room temperature (as shown in Figure 6). The device successfully retained its both resistance states for about half an hour.
Figure 6. Data retention performance of the ZnO device for 1800 seconds at room temperature.
So far, many models have been proposed to explain the resistive switching phenomenon. Among them all, the most widely reported phenomenon is based on formation of conducting filaments during set process and their annihilation during reset process. For ZnO based memory cells, the filamentary mechanism can also be utilized on the basis of current-voltage characteristics [13]. According to filament theory, defects might condense in the metal oxide to form tiny conducting filaments at HRS, and these tiny conducting filaments could gather and align to form long conducting filament, leading to transition to LRS [15]. In this work, conducting filaments in ZnO may be created along or with the walls of individual ZnO nanorod during set process. With applied potential reversely, these filaments can be ruptured easily to transform the device conductance state to its inherent phase. Thus a transition from low resistance state to high resistance state is observed.
In this report, a facile and cost-effective bottom up strategy of electrochemical deposition is deployed to fabricate ZnO thin films at various temperatures. Our findings suggest that samples prepared at 75 °C demonstrated greater crystallinity and well uniform surface morphology of ZnO nanorods than the samples prepared at low temperatures. Furthermore, ZnO samples prepared at 75 °C (ZnO nanorods) expressed superior resistive switching characteristics than other samples prepared at lower temperatures. Also, the ZnO nanorods based memory cells demonstrated a capability to retain its resistance states for 2000 switching cycles. Defects related switching mechanism was also proposed to support the results. The present work demonstrates that ZnO films have the potential for next generation non-volatile memory applications.
This research was sponsored by Australian Research Council (ARC) Discovery Projects of DP140104373, DP150103006.
The author declares no conflicts of interest in this paper.
| [1] | Y. Chen, C. Liu, W. Huang, S. Cheng, R. Arcucci, Z. Xiong, Generative text-guided 3d vision-language pretraining for unified medical image segmentation, preprint, arXiv: 2306.04811. https://doi.org/10.48550/arXiv.2306.04811 |
| [2] | Z. Wan, C. Liu, M. Zhang, J. Fu, B. Wang, S. Cheng, et al., Med-unic: Unifying cross-lingual medical vision-language pre-training by diminishing bias, preprint, arXiv: 2305.19894. https://doi.org/10.48550/arXiv.2305.19894 |
| [3] | C. Liu, S. Cheng, C. Chen, M. Qiao, W. Zhang, A. Shah, et al., M-FLAG: medical vision-language pre-training with frozen language models and latent space geometry optimization, preprint, arXiv: 2307.08347. https://doi.org/10.48550/arXiv.2307.08347 |
| [4] |
Z. Guo, K. Yu, N. Kumar, W. Wei, S. Mumtaz, M. Guizani, Deep distributed learning-based poi recommendation under mobile edge networks, IEEE Internet Things J., 10 (2023), 303–317. https://doi.org/10.1109/JIOT.2022.3202628 doi: 10.1109/JIOT.2022.3202628
|
| [5] |
Y. Jin, L. Hou, Y. Chen, A time series transformer based method for the rotating machinery fault diagnosis, Neurocomputing, 494 (2022), 379–395. https://doi.org/10.1016/j.neucom.2022.04.111 doi: 10.1016/j.neucom.2022.04.111
|
| [6] |
Q. Li, L. Liu, Z. Guo, P. Vijayakumar, F. Taghizadeh-Hesary, K. Yu, Smart assessment and forecasting framework for healthy development index in urban cities, Cities, 131 (2022), 103971. https://doi.org/10.1016/j.cities.2022.103971 doi: 10.1016/j.cities.2022.103971
|
| [7] |
J. Zhang, X. Liu, W. Liao, X. Li, Deep-learning generation of poi data with scene images, ISPRS J. Photogramm. Remote Sens., 188 (2022), 201–219. https://doi.org/10.1016/j.isprsjprs.2022.04.004 doi: 10.1016/j.isprsjprs.2022.04.004
|
| [8] |
Z. Guo, Y. Shen, S. Wan, W. Shang, K. Yu, Hybrid intelligence-driven medical image recognition for remote patient diagnosis in internet of medical things, IEEE J. Biomed. Health. Inf., 26 (2022), 5817–5828. https://doi.org/10.1109/JBHI.2021.3139541 doi: 10.1109/JBHI.2021.3139541
|
| [9] |
D. Zhang, X. Gao, A digital twin dosing system for iron reverse flotation, J. Manuf. Syst., 63 (2022), 238–249. https://doi.org/10.1016/j.jmsy.2022.03.006 doi: 10.1016/j.jmsy.2022.03.006
|
| [10] |
Z. Guo, Q. Zhang, F. Ding, X. Zhu, K. Yu, A novel fake news detection model for context of mixed languages through multiscale transformer, IEEE Trans. Comput. Social Syst., 2023. https://doi.org/10.1109/TCSS.2023.3298480 doi: 10.1109/TCSS.2023.3298480
|
| [11] |
X. Sun, Y. Zou, S. Wang, H. Su, B. Guan, A parallel network utilizing local features and global representations for segmentation of surgical instruments, Int. J. Comput. Assisted Radiol. Surg., 17 (2022), 1903–1913. https://doi.org/10.1007/s11548-022-02687-z doi: 10.1007/s11548-022-02687-z
|
| [12] |
Z. Chen, J. Chen, S. Liu, Y. Feng, S. He, E. Xu, Multi-channel calibrated transformer with shifted windows for few-shot fault diagnosis under sharp speed variation, ISA Trans., 131 (2022), 501–515. https://doi.org/10.1016/j.isatra.2022.04.043 doi: 10.1016/j.isatra.2022.04.043
|
| [13] |
M. Sun, L. Xu, R. Luo, Y. Lu, W. Jia, Ghformer-net: Towards more accurate small green apple/begonia fruit detection in the nighttime, J. King Saud Univ.-Comput. Inf. Sci., 34 (2022), 4421–4432. https://doi.org/10.1016/j.jksuci.2022.05.005 doi: 10.1016/j.jksuci.2022.05.005
|
| [14] | D. Chen, J. Zheng, G. Wei, F. Pan, Extracting predictive representations from hundreds of millions of molecules, J. Phys. Chem. Lett., 12 (2021), 10793–10801. |
| [15] |
N. P. Tigga, S. Garg, Efficacy of novel attention-based gated recurrent units transformer for depression detection using electroencephalogram signals, Health Inf. Sci. Syst., 11 (2023). https://doi.org/10.1007/s13755-022-00205-8 doi: 10.1007/s13755-022-00205-8
|
| [16] |
B. Wang, Q. Li, Z. You, Self-supervised learning based transformer and convolution hybrid network for one-shot organ segmentation, Neurocomputing, 527 (2023). https://doi.org/10.1016/j.neucom.2022.12.028 doi: 10.1016/j.neucom.2022.12.028
|
| [17] |
S. Xiao, S. Wang, Z. Huang, Y. Wang, H. Jiang, Two-stream transformer network for sensor-based human activity recognition, Neurocomputing, 512 (2022), 253–268. https://doi.org/10.1016/j.neucom.2022.09.099 doi: 10.1016/j.neucom.2022.09.099
|
| [18] | M. Mao, R. Zhang, H. Zheng, T. Ma, Y. Peng, E. Ding, et al., Dual-stream network for visual recognition, Adv. Neural Inf. Process. Syst., 34 (2021), 25346–25358. |
| [19] |
R. Kozik, M. Pawlicki, M. Chorś, A new method of hybrid time window embedding with transformer-based traffic data classification in iot-networked environment, Pattern Anal. Appl., 24 (2021), 1441–1449. https://doi.org/10.1007/s10044-021-00980-2 doi: 10.1007/s10044-021-00980-2
|
| [20] | A. A. Khan, R. Jahangir, R. Alroobaea, S. Y. Alyahyan, A. H. Almulhi, M. Alsafyani, et al., An efficient text-independent speaker identification using feature fusion and transformer model, Comput. Mater. Contin., 75 (2023), 4085–4100. |
| [21] |
D. Li, B. Li, S. Long, H. Feng, T. Xi, S. Kang, et al., Rice seedling row detection based on morphological anchor points of rice stems, Biosyst. Eng., 226 (2023), 71–85. https://doi.org/10.1016/j.biosystemseng.2022.12.012 doi: 10.1016/j.biosystemseng.2022.12.012
|
| [22] |
Y. Yang, J. Yu, H. Jiang, W. Han, J. Zhang, W. Jiang, A contrastive triplet network for automatic chest x-ray reporting, Neurocomputing, 502 (2022), 71–83. https://doi.org/10.1016/j.neucom.2022.06.063 doi: 10.1016/j.neucom.2022.06.063
|
| [23] |
B. Zhang, J. Abbing, A. Ghanem, D. Fer, J. Barker, R. Abukhalil, et al., Towards accurate surgical workflow recognition with convolutional networks and transformers, Comput. Methods Biomech. Biomed. Eng.: Imaging Visualization, 10 (2022), 349–356. https://doi.org/10.1080/21681163.2021.2002191 doi: 10.1080/21681163.2021.2002191
|
| [24] | X. Pan, X. Gao, H. Wang, W. Zhang, Y. Mu, X. He, Temporal-based swin transformer network for workflow recognition of surgical video, Int. J. Comput. Assisted Radiol. Surg., 18 (2023), 139–147. |
| [25] |
Y. J. Shin, S. B. Jeong, H. I. Seo, W. Y. Kim, D. H. Seo, A study on handwritten parcel delivery invoice understanding model, J. Adv. Mar. Eng. Technol. (JAMET), 46 (2022), 430–438. https://doi.org/10.5916/jamet.2022.46.6.430 doi: 10.5916/jamet.2022.46.6.430
|
| [26] |
Y. Liu, T. Bai, Y. Tian, Y. Wang, J. Wang, X. Wang, et al., Segdq: Segmentation assisted multi-object tracking with dynamic query-based transformers, Neurocomputing, 481 (2022), 91–101. https://doi.org/10.1016/j.neucom.2022.01.073 doi: 10.1016/j.neucom.2022.01.073
|
| [27] |
L. Tang, X. Xiang, H. Zhang, M. Gong, J. Ma, Divfusion: Darkness-free infrared and visible image fusion, Inf. Fusion, 91 (2023), 477–493. https://doi.org/10.1016/j.inffus.2022.10.034 doi: 10.1016/j.inffus.2022.10.034
|
| [28] |
H. Jiang, M. Gao, H. Li, R. Jin, H. Miao, J. Liu, Multi-learner based deep meta-learning for few-shot medical image classification, IEEE J. Biomed. Health Inf., 27 (2023), 17–28. https://doi.org/10.1109/JBHI.2022.3215147 doi: 10.1109/JBHI.2022.3215147
|
| [29] |
M. Luo, H. Wu, H. Huang, W. He, R. He, Memory-modulated transformer network for heterogeneous face recognition, IEEE Trans. Inf. Forensics Secur., 17 (2022), 2095–2109. https://doi.org/10.1109/TIFS.2022.3177960 doi: 10.1109/TIFS.2022.3177960
|
| [30] |
J. Izquierdo-Domenech, J. Linares-Pellicer, J. Orta-Lopez, Towards achieving a high degree of situational awareness and multimodal interaction with ar and semantic ai in industrial applications, Multimedia Tools Appl., 82 (2023), 15875–15901. https://doi.org/10.1007/s11042-022-13803-1 doi: 10.1007/s11042-022-13803-1
|
| [31] | Z. Yu, Y. Shen, J. Shi, H. Zhao, Y. Cui, J. Zhang, et al., Physformer++: Facial video-based physiological measurement with slowfast temporal difference transformer, Int. J. Comput. Vision, 131 (2023), 1307–1330. |
| [32] |
H. Ji, X. Cui, W. Ren, L. Liu, W. Wang, Visual inspection for transformer insulation defects by a patrol robot fish based on deep learning, IET Sci. Meas. Technol., 15 (2021), 606–618. https://doi.org/10.1049/smt2.12062 doi: 10.1049/smt2.12062
|
| [33] |
Y. Wu, K. Liao, J. Chen, J. Wang, D. Z. Chen, H. Gao, et al., D-former: A u-shaped dilated transformer for 3d medical image segmentation, Neural Comput. Appl., 35 (2023), 1931–1944. https://doi.org/10.1007/s00521-022-07859-1 doi: 10.1007/s00521-022-07859-1
|
| [34] | C. Liu, Z. Mao, A. Liu, T. Zhang, B. Wang, Y. Zhang, Focus your attention: A bidirectional focal attention network for image-text matching, in Proceedings of the 27th ACM International Conference on Multimedia, ACM, (2019), 3–11. https://doi.org/10.1145/3343031.3350869 |
| [35] | C. Liu, Z. Mao, T. Zhang, H. Xie, B. Wang, Y. Zhang, Graph structured network for image-text matching, in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, (2020), 10918–10927. https://doi.org/10.1109/CVPR42600.2020.01093 |
| [36] | H. Diao, Y. Zhang, L. Ma, H. Lu, Similarity reasoning and filtration for image-text matching, in Proceedings of the AAAI Conference on Artificial Intelligence, 35 (2021), 1218–1226. https://doi.org/10.1609/aaai.v35i2.16209 |
| 1. | Somnath S. Kundale, Akhilesh P. Patil, Snehal L. Patil, Prashant B. Patil, Rajanish K. Kamat, Deok-kee Kim, Tae Geun Kim, Tukaram D. Dongale, Effects of switching layer morphology on resistive switching behavior: A case study of electrochemically synthesized mixed-phase copper oxide memristive devices, 2022, 27, 23529407, 101460, 10.1016/j.apmt.2022.101460 | |
| 2. | Somnath S. Kundale, Girish U. Kamble, Pradnya P. Patil, Snehal L. Patil, Kasturi A. Rokade, Atul C. Khot, Kiran A. Nirmal, Rajanish K. Kamat, Kyeong Heon Kim, Ho-Myoung An, Tukaram D. Dongale, Tae Geun Kim, Review of Electrochemically Synthesized Resistive Switching Devices: Memory Storage, Neuromorphic Computing, and Sensing Applications, 2023, 13, 2079-4991, 1879, 10.3390/nano13121879 | |
| 3. | Pradnya P. Patil, Somnath S. Kundale, Shubham V. Patil, Santosh S. Sutar, Junseong Bae, Sunil J. Kadam, Krantiveer V. More, Prashant B. Patil, Rajanish K. Kamat, Seunghyun Lee, Tukaram D. Dongale, Self‐Assembled Lanthanum Oxide Nanoflakes by Electrodeposition Technique for Resistive Switching Memory and Artificial Synaptic Devices, 2023, 1613-6810, 10.1002/smll.202303862 |
Tao Wang, Min Qiu. A visual transformer-based smart textual extraction method for financial invoices[J]. Mathematical Biosciences and Engineering, 2023, 20(10): 18630-18649. doi: 10.3934/mbe.2023826
DownLoad: 
