Icon

01_​Data_​Preparation_​and_​CNN_​Training

Image Recognition for Retail Use Case: Data Preparation & Building Convolutional Neural Network (CNN) Workflow

We used neural networks for the image recognition task. Neural networks are massively parallel adaptive processing structures consisting of one or more layers, each layer one or more neurons. Three types of layers exist: Input (receiving data from its environment, providing processed data to other layers), hidden (receiving and providing processed data from and to other layers) and output (receiving processed data from other layers, providing information to the environment) layers. Weighted connections exist between the neurons of each layer, changing them is the key to its adaptability, which happens based on the difference between the predicted and the expected results.

Image Recognition for Retail Use CaseData Preparation & Building Convolutional Neural Network (CNN) WorkflowWe used neural networks for the image recognition task. Neural networks are massively parallel adaptiveprocessing structures consisting of one or more layers, each layer one or more neurons. Three types oflayers exist: Input (receiving data from its environment, providing processed data to other layers), hidden(receiving and providing processed data from and to other layers) and output (receiving processed data fromother layers, providing information to the environment) layers. Weighted connections exist between theneurons of each layer, changing them is the key to its adaptability, which happens based on the differencebetween the predicted and the expected results.A key advantage of neural networks is the fact that they can learn nonlinear relationships between variables.This gave researchers the idea to study neural networks in image recognition and object detection tasks. Reading Images & Image Preprocessing Partitoning & One hot encoding Training Model Evaluating Model DatasetWe do not provide the full training dataset as part of the worklfow to limit storage requirements. Change the path in the Image Preprocessing Wrapped Metanode to use a different data set for training.We would like to say thank you to the researchers of Freiburg University, who created and agreed to theusage of this dataset. You can download the dataset and documentation from here:https://github.com/PhilJd/freiburg_groceries_datasetYou can kindly reach the academic paper "The Freiburg Groceries Dataset" related this dataset via followingurl:https://arxiv.org/pdf/1611.05799.pdf Required Environment and Required KNIME ExtensionsTo run this workflow, you have to have Python 3 environment and Keras library on your local machine.Before running the workflow, following extensions of KNIME should be installed properly:- KNIME Deep Learning - Keras Integration (Labs)- KNIME Image Processing (Community Contributions Trusted)- KNIME Image Processing - Deep Learning Extension (Community Contributions Trusted) Getting pedicted classesConfusion Matrix and Predictions Image Preprocessing Format results Performance Metrics Preparationfor Training Training ConvolutionalNeural Network (CNN) Image Recognition for Retail Use CaseData Preparation & Building Convolutional Neural Network (CNN) WorkflowWe used neural networks for the image recognition task. Neural networks are massively parallel adaptiveprocessing structures consisting of one or more layers, each layer one or more neurons. Three types oflayers exist: Input (receiving data from its environment, providing processed data to other layers), hidden(receiving and providing processed data from and to other layers) and output (receiving processed data fromother layers, providing information to the environment) layers. Weighted connections exist between theneurons of each layer, changing them is the key to its adaptability, which happens based on the differencebetween the predicted and the expected results.A key advantage of neural networks is the fact that they can learn nonlinear relationships between variables.This gave researchers the idea to study neural networks in image recognition and object detection tasks. Reading Images & Image Preprocessing Partitoning & One hot encoding Training Model Evaluating Model DatasetWe do not provide the full training dataset as part of the worklfow to limit storage requirements. Change the path in the Image Preprocessing Wrapped Metanode to use a different data set for training.We would like to say thank you to the researchers of Freiburg University, who created and agreed to theusage of this dataset. You can download the dataset and documentation from here:https://github.com/PhilJd/freiburg_groceries_datasetYou can kindly reach the academic paper "The Freiburg Groceries Dataset" related this dataset via followingurl:https://arxiv.org/pdf/1611.05799.pdf Required Environment and Required KNIME ExtensionsTo run this workflow, you have to have Python 3 environment and Keras library on your local machine.Before running the workflow, following extensions of KNIME should be installed properly:- KNIME Deep Learning - Keras Integration (Labs)- KNIME Image Processing (Community Contributions Trusted)- KNIME Image Processing - Deep Learning Extension (Community Contributions Trusted) Getting pedicted classesConfusion Matrix and Predictions Image Preprocessing Format results Performance Metrics Preparationfor Training Training ConvolutionalNeural Network (CNN)

Nodes

Extensions

Links

Download

To use this workflow in KNIME, download it from the below URL and open it in KNIME:

Download Workflow
Created by: Olgun_Aydin
Created at: 2018-06-13
On NodePit since: 2019-07-12
Last update: 2023-12-05
Created with KNIME version: v3.6.0

Deploy, schedule, execute, and monitor your KNIME workflows locally, in the cloud or on-premises – with our brand new NodePit Runner.

Try NodePit Runner!