RoboFlow: HELP

RoboFlow is a semi-automated TensorFlow image classification explorer in a python command line app.

RoboFlow: HELP

jump to: RoboFlow | Help | DetailedHelp

This Roboflow guide assumes you have at least passing familiarity with TensorFlow:


Guided usage:


Advanced usage:

python [basetag] [imagequantity] [searchtag] [optional flowsteps]

for example:

python robots 200 robotart retrain


  • basetag: the master classification tag or theme (‘robots’ or ‘birds’ or whatever)
  • imagequantity: an integer between 0 and the {imgnum_maxpercycle} var from the config file, to not hammer other people’s servers
  • searchtag: the searchterm, such as ‘robot’ or ‘robotart’ etc
  • flowsteps: OPTIONAL parameter to determine which stages to implement of –
    ** ‘download’ - only downloads the tagged images
    ** ‘classify’ (default, if blank) - download, classify/sort images w/model of your choice
    ** ‘classify_top’ - download, classify with TOP accuracy model (skips setup),
    ** ‘retrain’ - download, classify, and retrain (w/optional harvest) a new
    classifier with images from training_photos/{basetag}
    ** ‘retrain_defaults’ (as in ‘robots 0 0 retrain_defaults’) will skip the retrain SETUP,
    using values from config file setup. ** ‘automatic’ - combo of ‘classify_top + retrain_defaults’ in one parameter


Quality initial labeling/sorting makes ALLL the difference!
When starting, you must manually sort a minimum number of images to allow the first retraining to create the first classifier. After that, downloading, classifying/auto-sorting, and harvesting to retrain to make a better classifier is waaaaay more automated.

Keeping reading the extensive detailed help for useful and important advice.