search

Alibaba-MIIL / ASL

Official Pytorch Implementation of: "Asymmetric Loss For Multi-Label Classification"(ICCV, 2021) paper
MIT License
732 stars 102 forks source link

Check map of OpenImagesV6 with TResNet_L Backbone after convert model to onnx #75

Closed Chenchienyuchen closed 2 years ago

Chenchienyuchen commented 2 years ago

I want to check OpenImagesV6 map value with TResNet_L of onnx.

I check validate.py is for Coco Dataset. Could you provide validate script for OpenImagesV6 dataset ? Or can you give me some suggestion? Thanks a lot.

mrT23 commented 2 years ago

i am not gonna provide a dedicated script for open-images. In any case open-images dataset is very hard to download.

my recommendation is just to run inference on several images, and see that results are logical and informative, and are the same for onnx and pytorch files.

p.s there is a detailed inference example in the README file for open-images, follow it

p.s.2 better open-images models can be found at: https://github.com/Alibaba-MIIL/PartialLabelingCSL