yanxp / MetaR-CNN

Meta R-CNN : Towards General Solver for Instance-level Low-shot Learning
https://yanxp.github.io/metarcnn.html
179 stars 23 forks source link

How was the test process going? #49

Open yang-yk opened 4 years ago

yang-yk commented 4 years ago

Hi, @yanxp.

Sorry to bother you. I am very interested in your work named Meta-RCNN. In the training process, you have provided three data splits to train the meta-model in VOC dataset. However, in the test process, you also need a support set and a query set to test the model. I wonder how you get the support set and query set in novel classes of VOC dataset for 1/2/3/5/10-shot experiments. And how many episodes are there when you test the model. Can you provide me a specific file to illustrate the support set and query set of VOC dataset for 1/2/3/5/10-shot experiments like the data splits in the training process? By the way, you have made it clear that you use resnet-101 as the backbone for 3/10-shot experiments. Can you tell that me whether you use resnet-101 as the backbone for the 1/2/5-shot experiment in VOC dataset. Many thanks and hope to hear from you soon.

Best Regard Yukuan Yang