After training my dataset for 13000 iterations, the training, validation, and testing sets all show 0% accuracy and all my test photos show false negative/no object detected. The dataset has 1032 photos and 2 classes, and I used Roboflow for the image annotation. If there is any way to fix this?
Here is a photo of my project, I tested this one with 100 iterations instead of 13000 but it still produced 0%. CreateML project photo
I used Roboflow for annotating the images and then exported the dataset to CreateML in download zip code format, and inserted the train, valid, and testing photos into Createml. I chose Full network and 13000 iterations with 13 x 13 grid and pressed train. After a day of training the lost was about 0.0094 but the train, valid, and testing sets all show 0%, and in the evaluation, the testing dataset showed 0% accuracy with all photos being false negative.