JayHeYang / pedestrian-attributes-recognition

pedestrian-attributes-recognition

Introdution

A pytorch implemented for pedestrian attributes recognition base.

This is a basic engineering code, you can add or delete some content according to your own needs, so as to achieve the appearance you want.

For example, to increase the network structure, you need to add network files in src/models; to modify the option parameters, you can modify them in options/options.

Directory

• dataset

The data set file, which contains image path and labels of attributes.

test.png 1 0 1

• model

  The model of pre-trained or want to store.

• image

  The image files.

• model

  The model of pre-trained or want to store.

• output

  The log file in train or test process.

• src

• data: load data.
• options: parameters.
• loss: loss functions.
• models: networks.
• util: utils.
• train.py: train model.
• deploy.py: test model.

Train

You should modify the options.py firstly, and then:

python train.py

Test

You should modify the options.py firstly, and then:

python deploy.py

Reference

• pytorch-multi-label-classifier