Error after trying to train on custom dataset

xg-chu / CrowdDet

[CVPR 2020] Detection in Crowded Scenes: One Proposal, Multiple Predictions

MIT License

422 stars 85 forks source link

Error after trying to train on custom dataset #59

Open sherry30 opened 3 years ago

sherry30 commented 3 years ago

Hello! Getting this error in softmaxLoss function after chaning the number of classes from 2 to 13 to train on custom dataset Screenshot (15)

classes: class_names = ['bridge', 'airplane','ground track field','vehicle','parking lot','T junction','baseball diamond','tennis court','basketball court','ship','crossroad','harbor','storage tank']

I have only made changes in the config file so far, setting up paths and class_names of my dataset

yexiguafuqihao commented 2 years ago

The number of classes needs to be equal to your customized dataset.