Closed Sarthak-Shah closed 5 years ago
What classes of objects do you want to predict? What prediction results are you getting from your custom trained model?
Hi , I want to predict one Android application's typical icons whch used to slightly alter according to different Android UI and resolution. So I am developing one class from which my trained model will detect that icon and click it by finding coordinate further using UIautomator.
On Thu, May 30, 2019, 7:22 PM MOSES OLAFENWA notifications@github.com<mailto:notifications@github.com> wrote:
What classes of objects do you want to predict? What prediction results are you getting from your custom trained model?
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/OlafenwaMoses/ImageAI/issues/221?email_source=notifications&email_token=AEBMI7HZNCRI33Q2YWKW2FDPX7LX7A5CNFSM4HQNJDX2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODWSMDSY#issuecomment-497336779, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AEBMI7AQE3QKCY4JVI24JX3PX7LX7ANCNFSM4HQNJDXQ.
Please refer to #8 for object detection
Though i have trained my model and given the proper path to locate .JSON and .h5 files after model training, it's not revoking predictions and probabilities as per new model. somehow its playing in Resnet model only and using that predefined person's designation only.