search

dk-liang / FIDTM

[IEEE TMM] Focal Inverse Distance Transform Maps for Crowd Localization
MIT License
169 stars 41 forks source link

Training on custom dataset #37

Closed mariosconsta closed 1 year ago

mariosconsta commented 1 year ago

Hello

If I want to train this model on my own custom dataset, I just need to change the files _fidt_generate_xx.py, make_npydata.py and train_baseline.py_?

Also, my dataset contains bounding boxes, but that shouldn't be a big issue right?

dk-liang commented 1 year ago

Yes, you are right

mariosconsta commented 1 year ago

Yes, you are right

Thank you for the swift response. I am trying to solve the low performance on crowd localization where the view is top-down from a drone. Since it's top down, only the heads are visible as blobs, and I think if I train it on some of my data which is only top-down, it might increase its performance.