Hey guys I'm a medical student researcher doing some simple work using CreateML and iOS. Im using CreateML and made a model using the object detection, when I get to the evaluation section I’m lost, traditionally for image classification it tells you the precision/accuracy of the model based on its classes but for Object detection it’s different. The section I was confused on was the evaluation/accuracy tab where the Object detection model talks about I/U 50% and Varied I/U and gives us %s for each of those classes. Could you explain to me what the I/U 50% and the Varied I/U represent? I tried searching across the web but almost nobody references those specifically and discusses what they mean. I dont have much of an AI/ML background so its a foreign concept to me. I would greatly appreciate it! Thank you!
CreateML object detection evaluation
Hey guys I'm a medical student researcher doing some simple work using CreateML and iOS. Im using CreateML and made a model using the object detection, when I get to the evaluation section I’m lost, traditionally for image classification it tells you the precision/accuracy of the model based on its classes but for Object detection it’s different. The section I was confused on was the evaluation/accuracy tab where the Object detection model talks about I/U 50% and Varied I/U and gives us %s for each of those classes. Could you explain to me what the I/U 50% and the Varied I/U represent? I tried searching across the web but almost nobody references those specifically and discusses what they mean. I dont have much of an AI/ML background so its a foreign concept to me. I would greatly appreciate it! Thank you!
Add a Comment
