Open Junhyuk opened 6 years ago
I think you can try some open database, such as MS COCO or VOC.
@tianyuecao Is it necessary to modify the code with this, the MS COCO format and the data set of the project are not the same?I look forward to hearing from you, thank you in advance.
@tianyuecao Is it necessary to modify the code with this, the MS COCO format and the data set of the project are not the same?I look forward to hearing from you, thank you in advance.
Not exactly. I'm now using the objects in each image as the labels, thus I have to extract it from the annotation file. But i'm now using torch to implement it, maybe i'll write a dataloader for MS COCO some time :)
@tianyuecao What is the accuracy of the MS COCO dataset you tested? ? This project is a txt for each picture, then does it have a txt for the MS COCO data set?
@tianyuecao I am researching multi-label image classification, if you have the chance, we can add a WeChat exchange. upFF222
@tianyuecao What is the accuracy of the MS COCO dataset you tested? ? This project is a txt for each picture, then does it have a txt for the MS COCO data set?
Actually I didn't apply it on MS COCO. I've tested it on a dataset from my lab. You can extract txt files for MS COCO from its annotation files if you'd like to.
@tianyuecao I am researching multi-label image classification, if you have the chance, we can add a WeChat exchange. upFF222
I'd like to, maybe leave me an email?
@tianyuecao 173883436@qq.com
Please comment if you have them.