Jessespace

Any new ?

I have tried it in my own dataset,2 labels making by **labelImg** You should change the **.xml** file like the VID dataset.Then ,following the VID trainning steps.

> U can just make your datasets as the DET/VID format, or u can change the data loading codes..... I have tried and succeed. But honestly，the mAP is very slow,have...

> @Jessespace I also encountered the same problem. I want to train my data set with DFF, but I haven't downloaded ILSVRC2015 VID data set yet. I don't know what...

Did you solve it?

> I train the model from scratch using my own dataset, and the training process is very normal, but when use MAPMetric after each forward, the mAP is very low,...

> @Jessespace By the way, you can try gluon-cv, it's much convenient and powerful. Thanks,

> Random is good for training, and you may take frames from one file so many times that every frame shoud be taken times.So ,there is no data-wasting if you...

try with python2/anaconda2