TensorFlow-Object-Detection-API-Tutorial-Train-Multiple-Objects-Windows-10
TensorFlow-Object-Detection-API-Tutorial-Train-Multiple-Objects-Windows-10 copied to clipboard
EdjeElectronics
Invalid SOS parameters for sequential JPEG in training
I'm having this "error" when training is going on What could it be?
@Swordancer Thanks, I've had some very large images> = 3 thousand pixels wide after that decreases the size of the images stopped appearing this error, but now my parameter loss is giving a giant value, I do not know what can be.
Do you know what it could be?
I am trying to train Brazilian notes, there is a difficulty in training because there is a lot of similarity in the back and I do not know if it is a problem for the training.
How many of these pictures do you have? Do you have the same problem with other pictures or other zoo models? Do not use pictures with to big resolution. All pictures should be with similar quality. I these are your images, then I have no idea what's cuse the problem.
@Swordancer So I tested the animal classification model and it worked, I have several images of these, they are in total 6 classes of values and I have about 200 images in total, there is the other side also that has more unusual details.
Hi! I got the same problem. How did you solved it?
Did it gave some problems after the training? Or it doesn't matter and your model is working good??
Is this solved? I am facing the same issue. I see that the training and validation is finishing, but as soon as these finishes, i get this "Invalid SOS parameters for sequential JPEG"
I'm also facing this issue when training my object detection model, does anyone knows the solution?
