[ad_1]
I’m utilizing the Azure Customized Imaginative and prescient service customvision.ai for information labelling. I educated my information from Azure and it could actually detect the objects through API or through quicktesting from the customvision.ai. I exported it to CoreML by the insturctions from Microsoft’s documentation, however I could not make it work for ios or xcode.
There’s a preview tab for coreML recordsdata on xcode. The pattern fashions I downloaded from the web works as anticipated. Nonetheless, I can’t get the anticipated outcomes for my mannequin from xcode, in comparison with the API or quicktest through web site.
How can I get it work on my ios app. I’m utilizing Basic (compact) [S1] area for my mannequin.
Progress Replace:
I’ve discovered a pattern code from Azure’s github. It really works with their mannequin. Nonetheless once I change the mannequin and config recordsdata to my ones, it throws Dangerous entry error someplace.
Additionally I acknowledged the exporter variations are totally different and there’s an additional metadata_properties.json in my model. So even when the readme from the repo states any 2.x model is supported, (mine is 2.1 however their mannequin is 2.0) I suppose that pattern code does not assist the model I exported.
Can I downgrade my mannequin’s model or are there any pattern swift initiatives for the up to date model.
[ad_2]
