search

labyrinth7x / Deep-Cross-Modal-Projection-Learning-for-Image-Text-Matching

Deep Cross-Modal Projection Learning for Image-Text Matching
72 stars 21 forks source link

How to reproduce the performence on CUHK-PEDES? #10

Open liyingGao opened 4 years ago

liyingGao commented 4 years ago

I use python 3.6.4, Pytorch1.0.0 & torchvision 0.2.1, scipy 1.2.1. The results in paper 'Deep Cross-Modal Pojection Learning for Image-Text Matching' on CUHK-PEDES are:{top- 1 = 49.37%，top-10 = 79.27%}, but I only get {top- 1 = 38.35%，top-10 = 63.39%} using MobileNetv1 as backbone, and {top- 1 = 41.44%，top-10 = 65.66%} using Resnet152. I wander if anyone could reproduce the results, and if it is convenient, please share the training details and hypeparameters.

npumazehong commented 3 years ago

I use python 3.6.4, Pytorch1.0.0 & torchvision 0.2.1, scipy 1.2.1. The results in paper 'Deep Cross-Modal Pojection Learning for Image-Text Matching' on CUHK-PEDES are:{top- 1 = 49.37%，top-10 = 79.27%}, but I only get {top- 1 = 38.35%，top-10 = 63.39%} using MobileNetv1 as backbone, and {top- 1 = 41.44%，top-10 = 65.66%} using Resnet152. I wander if anyone could reproduce the results, and if it is convenient, please share the training details and hypeparameters. ? I have the similar results like you.Have you reproduced the paper's result using pytorch ? Can we share with each other?

yyll1998 commented 3 years ago

I use python 3.6.4, Pytorch1.0.0 & torchvision 0.2.1, scipy 1.2.1. The results in paper 'Deep Cross-Modal Pojection Learning for Image-Text Matching' on CUHK-PEDES are:{top- 1 = 49.37%，top-10 = 79.27%}, but I only get {top- 1 = 38.35%，top-10 = 63.39%} using MobileNetv1 as backbone, and {top- 1 = 41.44%，top-10 = 65.66%} using Resnet152. I wander if anyone could reproduce the results, and if it is convenient, please share the training details and hypeparameters. ? I have the similar results like you.Have you reproduced the paper's result using pytorch ? Can we share with each other?

I have the same problem with you.Have you solved this problem? If you have,could we share with each other?Thank you !