Deep Learning based Coffee Beans Quality Screening

A Remote Access Server with Chatbot User Interface for Coffee Grinder Burr Wear Level Assessment Based on Imaging Granule Analysis and Deep Learning Techniques

“…That is why the input vector may not be susceptible to small size errors in coffee granules in our application. However, if the applications need to obtain deterministic size values, such as the applications in [13][14][15][16][17][18][19][20], the simple scale calibration using a reference object of known size could not be accurate enough. One should use image acquisition systems with pixel to length scale calibration, ambient brightness control, and a fixed shooting distance.…”

“…The devices should be able to fix the shooting distance between the camera and the objects, control the ambient brightness during photography, and have a calibrate function to determine the scale of the image pixels to the actual length. For example, custom-built image-capturing platforms [13][14][15], microscopes with acquisition software [16][17][18][19], and high-resolution digital CCD industrial cameras with acquisition software [20] have brightness controls and known calibrated pixel to length ratios. In our case, for user convenience, by using a flash light, the coffee granule images are able to be taken by mobile phones at a distance roughly between 15 and 25 cm without precise pixel to length calibration.…”

“…CNN and its derived methods can directly use images as inputs to train the deep learning model for classification applications, and it has become very popular in coffee-related applications. For example, the CNN-based or derived models have been successfully applied in coffee bean inspection machines [38], coffee bean quality screening [13], coffee bean defects [39,40], coffee leaf disease classification using ResNet50 [41] and MobileNetV3 [42], coffee maturity classification [43], and roasting coffee bean quality assessment using Mo-bileNetV2 [44]. A good comparison of CNN-derived models is available in [44].…”

“…Although CNN-based methods using images as an input appear to be very effective and convenient to apply, annotations or labels are usually needed in the training and validation sets for object detection [26,45] and image classifications [13,[39][40][41][42]. A comprehensive review on image segmentation algorithms based on deep learning models with annotated datasets is also available in [32].…”

Deep learning-empowered crop breeding: intelligent, efficient and promising

“…Developing a model to estimate the number of leaves and plant age for watermelon plants, and classifying them under normal and low-temperature stress, facilitates growth monitoring and improves water and sugar content in watermelons ( Nabwire et al., 2022 ). Training deep learning models to classify coffee leaf images and determine if they are infected with leaf rust disease can aid in early detection of diseases and enables timely measures to protect coffee crop yields and quality ( Shao et al., 2022 ). Furthermore, deep learning can analyze the correlation between phenotypic genomic, facilitating precise selection and optimization of genomic combinations as well as gene editing techniques to improve crop yields and quality.…”