mmfashion_retrieval

MMFashion -- In-Shop Clothes Retrieval

Input

(Image from https://github.com/open-mmlab/mmfashion/blob/master/demo/imgs/06_1_front.jpg)

Shape : (1, 3, height, width)

Reshaped to : (1, 3, 224, 224)

Output

• Top-k results

Usage

Automatically downloads the onnx and prototxt files on the first run. It is necessary to be connected to the Internet while downloading.

For the sample image,

$ python3 mmfashion_retrieval.py

By specifying the path to the root folder of the dataset of images with the --gallery option, you can use your own gallery of images for the retrieval task.

$ python3 mmfashion_retrieval.py --gallery PATH_TO_THE_ROOT_FOLDER_OF_YOUR_GALLERY 

If you want to specify the input image, put the image path after the --input option.
You can use --savepath option to change the name of the output file to save.

$ python3 mmfashion_retrieval.py --input IMAGE_PATH --savepath SAVE_IMAGE_PATH --gallery PATH_TO_THE_ROOT_FOLDER_OF_YOUR_GALLERY

By adding the --video option, you can input the video.
If you pass 0 as an argument to VIDEO_PATH, you can use the webcam input instead of the video file.

$ python3 mmfashion_retrieval.py --video VIDEO_PATH --gallery PATH_TO_THE_ROOT_FOLDER_OF_YOUR_GALLERY

By specifying the -k option, you can choose how many images you want to retrieve from the gallery.

$ python3 mmfashion_retrieval.py -k 2 --gallery PATH_TO_THE_ROOT_FOLDER_OF_YOUR_GALLERY

Reference

• MMFashion

Framework

Pytorch 1.9.1

Python 3.7+

Model Format

ONNX opset=11

Netron

in-shop_retriever.onnx.prototxt