image-embedding-resnet50/README.md

Image Embedding Pipeline with Resnet50

Authors: derekdqc, shiyu22

Overview

This pipeline is used to extract the feature vector of the image. First step is to normalize the image, and then use resnet50 model to generate the vector.

The pipeline parses the yaml file, which declares some components we call Operator and DataFrame. Next, we will introduce the interface, show how to use it and how it works, have fun with it!

Interface

towhee.pipeline(task: str, fmc: FileManagerConfig = FileManagerConfig(), branch: str = 'main', force_download: bool = False) [source]

params:

• task(str): task name or pipeline repo name.
• fmc(FileManagerConfig): optional, file manager config for the local instance, default is a default FileManagerConfig obejct.
• branch(str): optional, which branch to use for operators/pipelines on hub, defaults to 'main'.
• force_download(bool): optional, whether to redownload pipeline and operators, default is False.

return:

• _PipelineWrapper, an instance of the wrapper class around Pipeline.

When we declare a pipeline object with a specific task, such as towhee/image-embedding-resnet50 in this repo, it will run according to the Yaml file, and the input and output are:

inputs:

• img_tensor(PIL.Image), the image to be encoded.

outputs:

• cnn(numpy.ndarray), the embedding of the image.

How to use

1. Install Towhee

$ pip3 install towhee

You can refer to Getting Started with Towhee for more details. If you have any questions, you can submit an issue to the towhee repository.

2. Run it with Towhee

>>> from towhee import pipeline
>>> from PIL import Image

>>> img = Image.open('path/to/your/image') # for example, './test_data/test.jpg'
>>> embedding_pipeline = pipeline('towhee/image-embedding-resnet50')
>>> embedding = embedding_pipeline(img)

How it works

First of all, you need to learn the pipeline and operator in Towhee architecture:

• Pipeline: A Pipeline is a single machine learning task that is composed of several operators. Operators are connected together internally via a directed acyclic graph.

• Operator: An Operator is a single node within a pipeline. It contains files (e.g. code, configs, models, etc...) and works for reusable operations (e.g., preprocessing an image, inference with a pretrained model).

This pipeline includes four functions: _start_op, towhee/transform-image, towhee/resnet50-image-embedding and _end_op. It is necessary to ensure that the input and output of the four Operators correspond to each other, and the input and output data types can be defined by DataFrame.

Among the four Operator, _start_op and _end_op are required in any Pipeline, and they are used to start and end the pipeline in the Towhee system. For the other two Operators, please refer to towhee/transform-image and towhee/resnet50-image-embedding.