Sorting apparatus of ginseng

The present invention provides a fresh ginseng sorting machine that learns images of fresh ginseng of irregular shapes using deep learning technology to identify colors, sizes, numbers of roots, etc., and classify the fresh ginseng into appropriate grades. The fresh ginseng sorting machine in accord...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: KIM, TAE HYEUNG, KANG, KUN MOOK
Format: Patent
Sprache:eng ; kor
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The present invention provides a fresh ginseng sorting machine that learns images of fresh ginseng of irregular shapes using deep learning technology to identify colors, sizes, numbers of roots, etc., and classify the fresh ginseng into appropriate grades. The fresh ginseng sorting machine in accordance with the present invention comprises: a photographing unit (10) installed on one side or an upper surface of a transfer sorting unit (30) to photograph moving fresh ginseng; a transfer sorting unit (30), which is a transfer device sorting harvested fresh ginseng, unloading fresh ginseng loaded on a seat into a corresponding classification box in accordance with a determination result; a control unit (20) controlling the photographing unit and the transfer sorting unit and classifying and unloading the fresh ginseng in accordance with a result of an image determination unit; a data learning unit (110) building a learning model with learning data sets collected by the photographing unit and verifying the learning model by testing the learning model with verification data; and an image determination unit (120) for determining by comparing and analyzing the learning model and an actual image taken, wherein the data learning unit involves a convolutional neural network and a recurrent neural network algorithm applied thereto. 본 발명에 따른 수삼선별기는, 이송선별부(30)의 일측 또는 상면에 설치되어 이동하는 수삼을 촬영하는 촬영부(10); 수확한 수삼을 선별하기 위하여 이송하는 장치로 안좌에 적재된 수삼을 판단 결과에 따라 해당 분류함으로 하적하는 이송선별부(30); 상기 촬영부 및 이송선별부를 제어하고 이미지 판단부의 결과에 따라 분류하여 수삼을 하적하도록 제어하는 제어부(20);상기 촬영부에 수집한 학습 데이터 셋으로 학습 모델을 구축하고 검증 데이터로 테스트를 하여 학습 모델을 검증하는 데이터 학습부(110) 및 상기 학습 모델과 촬영된 실제 이미지를 비교 분석하여 판단하는 이미지 판단부(120)를 포함하되, 상기 데이터 학습부는 합성곱 신경망과 순환 신경망 알고리즘이 적용된 것을 특징으로 한다.