I want to try my trained model at local.(same classified with diffrent data at lesson2) . I am opening image with open_image and using line of code : pred_class,pred_idx,outputs = learn.predict(im) .
Yes it is work but it is work wrong. Always my model predicts my images as same category. When ı try it on kaggle it works very well. İs there something wrong with versions of my local and kaggle ? Or is my model on overfitting?Or am I doing something wrong ? And my model has .pkl extension is it right ? Thanks.
Maybe share your code and we can have a look. Hard to tell why that might happen without seeing the code in detail. Could be a minor mistake.
learn=load_learner(Path('E:\İndirmeler'),fname=Path('export.pkl'))
im=cv2.imread('One.jpg')
t=pil2tensor(im,dtype=np.float32)
im=Image(t)
pred_class,pred_idx,outputs=learn.predict(im)
print(pred_class)
print(outputs)still same here