KMITL

Permanent URI for this communityhttps://dspace.kmitl.ac.th/handle/123456789/1

Browse

Settings

Subject: search.filters.subject.OCR

Search Results

Now showing 1 - 4 of 4
  • Loading...
    Thumbnail Image
    Some of the metrics are blocked by your 
    consent settings
    Item type:Publication,
    Three-stage deep learning system for recognizing contaminated serial numbers in hard disk drive: A comparison study with two-stage deep learning model
    (2022-01-01)
    Chousangsuntorn, Chousak
    ;
    Tongloy, Teerawat
    ;
    Chuwongin, Santhad
    ;
    Boonsang, Siridech
    The previous two-stage deep learning model for detecting and classifying misidentified serial numbers on the defect hard disk drive (HDD) slider was proposed by authors. We found that the threshold level adjusted during preprocessing process could limit the robustness of the two-stage model in large-scale manufacturing. Thus, we proposed a three-stage deep learning model comprised of 1) region of interest (ROI) detection and cropping, 2) character detection and cropping, and 3) character classification. Object detection algorithm and classification network used in this model are based on YOLO v.4 and EfficientNet-B0. The 1000 images captured by the digital camera were used for training (600 images) and validation (400 images) of the ROI detection model. The other 1000 captured images were used for testing the performance of the proposed three-stage model, then we compared them with those obtained from the previous two-stage model. The proposed three-stage model reaches F1 score = 0.997 and recovery rate up to 95.9%, while the two-stage model yields only 0.948 and 73%, respectively.
  • Loading...
    Thumbnail Image
    Some of the metrics are blocked by your 
    consent settings
    Item type:Publication,
    Runner BIB number recognition system
    (2017-12-13)
    Anuntachai, Anuntapat
    ;
    Chaorattana, Wanatphong
    ;
    Boonchoay, Jutatip
    This research represents the runner BIB number recognition system to develop image processing study which solves problems and increases efficiency about runner image management in running fairs. The runner BIB number recognition system processes runner image to recognize BIB number and time when runner appears in media. The information from processing has collected to applicative later. BIB number position is on BIB tag which attach on runner body. To recognize BIB number, the system detects runner position first. This process emphasize on runner face detection in images following to concept of researcher then find BIB number in body-thigh area of runner. The system recognizes BIB number from BIB tag which represents in media. This processing presents 0.80 in precision value, 0.81 in recall value and F-measure is 0.80. The results display the runner BIB number recognition system has developed with high efficiency and can be applied for runner online communities in actual situation. The runner BIB number recognition system decreases problems about runner image processing and increases comfortable for runners when find images from running fairs. Moreover, the system can be applied in commercial to increase benefits in running business.
  • Loading...
    Thumbnail Image
    Some of the metrics are blocked by your 
    consent settings
    Item type:Publication,
    Japanese ingredient facts recognition application
    (2017-11-03)
    Tasanaprasert, Supawan
    ;
    Putchakarn, Vasinee
    ;
    Auerach, Patcharaporn
    Specific ingredients cautious people have difficulties in choosing products which written in Japanese language. Especially, people with the allergy has to be very conscious because wrong in take can cause death or allergic reaction. Thus we aim to develop Japanese ingredient facts recognition application. The process of this application goes as follow: first scan ingredient label of product and then application can detect individual character and match with its database towards specific items. Finally user will be warned if the specific items are used. The accuracy of this application relies on the clarity of the scan.
  • Loading...
    Thumbnail Image
    Some of the metrics are blocked by your 
    consent settings
    Item type:Publication,
    3D metallic surface acquisition for inspection and reading character
    (2013-11-04)
    Suesut, T.
    ;
    Gulphanich, S.
    ;
    Roonprasang, K.
    This paper presents a real-time implementation of 3D acquisition for reading text and inspection the metallic surface based on light sectioning. A measurement is achieved with a standard low cost CMOS camera. Surface defects are modeled as deviations in the local relief from a smooth approximation of the surface. Discrete orthogonal bases are used to generate a smoothed global model of the surface structure. Modified discrete Tchebychev polynomials are used as orthogonal basis functions to perform least square approximations of the geometry. QR decomposition is used to obtain a unitary basis, minimizing the numerical effort when modeling surfaces. The result of test measurements on copper sheets in a production environment is presented to demonstrate the surface inspection. Another result is shown the readable character on the textured metallic surface after the proposed processing. A prototype system of the laser scanning instrument can be implemented in a production line as well. © (2013) Trans Tech Publications, Switzerland.