11/15/2022, 8:44 AM
Hello, I have a question regarding packaging/using a kedro project. I would like to make a kedro pipeline available to evaluate new data. For example I have used kedro to train and evaluate a neural network on a set of images and I would like to create a new pipeline and deliver it to predict on some new images. I can well see how to construct the pipeline (which is composed of multiple image transformation using kedro parameters), but I don't know how to specify an arbitrary input. Ideally, I would copy my data folder to provide the trained model to the script collecting an image (through api, camera, etc...), import my package created with kedro and finally provide the collected image as the input of the pipeline which would then return a prediction. Is it possible or am I trying to abuse kedro with an application it was not planned for?