So I am curious... when I invoke the API (using python at the moment) as follows:
image = ClImage(file_obj=open('/Volumes/MEDIA/blah/blah/2017/2017-08-20/IMG_1111.JPG', 'rb'))
result = model.predict([image])
print json.dumps(result, indent=4, sort_keys=True)
what actually happens to that image file? I assume that the image, or some downsampled version of it, is sent to Clarifai's server for analysis. does a copy of my image remain on that server? for how long? in other words, am I handing over my entire photo archive to Clarifai if I use the API to generate predictions for my images?
I also assume (haven't read all the doco yet so forgive me if this is an obvious rtfm) that I don't really need to send full rez. what's the lowest rez that Clarifai considers adequate for analyzing an image?
I'm mostly interested in the JSON output (to feed it into my own code for auto-keywording images with simple tags like "sunset," "boat," "mountain," "bird," etc). I'd like to explore training my own model, eventually. but I'd also like to retain control over my image library (I'm not a pro but I do sell some prints occasionally, so for both privacy and commercial reasons I'd like to know who has access to my photos). should I be worried that by using Clarifai I may be losing ownership of my images?