- If I am not using the template and if I want to upload my dataset according to this tutorial: https://github.com/EdjeElectronics/TensorFlow-Object-Detection-API-Tutorial-Train-Multiple-Objects-Windows-10, what am I supposed to upload in floydhub?
Both the code and the dataset.
2 . In which terminal do I change the path of the slim folder?
Inside the FloydHub machine Terminal.
3 . What is the command to train the tf model?
It depends from your use-case but usually it something like this:
python train.py --logtostderr --train_dir=training/ --pipeline_config_path=training/faster_rcnn_inception_v2_pets.config
4 . Are you all running the commands from the floydhub terminal or through your command prompt?
It depends on which system you have to work.
5 . Did you upload the entire folder of object_detection from your system to floydhub as a dataset or did you just upload the train folder, test folder, train.record and test.record?
To run the TF Object Detection code you need to upload all the
object_dection folder, install/upload the slim package and upload your dataset as a FloydHub dataset.
6 . If you use Anaconda prompt, and run the floyd run ... command, do you guys mount the data to some local directory in the system? How to mount the dataset? What is the meaning of mount point? Does it mount the data set to some local mount point in the system's directory or does it mount to some floyhub's directory?
Take a look at these pages from the docs:
Hope that helps.