You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am trying to infer the gesture of test images present in ./imgs folder by loading the CNN with you pretrained weights file pretrained_weights_WinOS.hdf5 but the prediction is going wrong. Like for pppunch1.png image it is predicting PEACE gesture. Can you tell me where is it going wrong. Below is the screenshot of my output logs.
The text was updated successfully, but these errors were encountered:
Hey @SumanDeedwaniya
Yeah there could be few false positives. I am using a very simplistic model, trained with a very limited set of images and its not 100% accurate model. I must have copied that image file from the validation set.
To improve the accuracy, you may need to train it with more training data. Few model tweaking may also help.
@sonalgan Hey! Always welcome to contribute to this project. Appreciate any help.
Some of the new additions would be:
support for latest Keras/Tensorflow versions.
commandline interface could be quick update using python 'argeparse' lib.
I was working on containerize the project using Docker framework. But hit a blocking issue. I want to host the project as well online for someone to quickly use and interface it.
Hi,
I am trying to infer the gesture of test images present in ./imgs folder by loading the CNN with you pretrained weights file pretrained_weights_WinOS.hdf5 but the prediction is going wrong. Like for pppunch1.png image it is predicting PEACE gesture. Can you tell me where is it going wrong. Below is the screenshot of my output logs.
The text was updated successfully, but these errors were encountered: