Arnaldo Gualberto
Arnaldo Gualberto
Hi, @pkarimib take a look at my [Medium article][medium] about how I built the dataset, please. [medium]: https://medium.com/analytics-vidhya/face-pose-estimation-with-deep-learning-eebd0e62dbaf
take a look in this link: [https://we.tl/t-Y1RedLHx3v](https://we.tl/t-Y1RedLHx3v), @D1vyansh
I have code for running TensorFlow in C++ only for TF v1.4. Certainly, it does not work for v2.x.
I used [this code](https://gist.github.com/arnaldog12/5c6494d20a4a5d7b23a01975b66811b1) to run TensorFlow models in C++ using OpenCV to work with images.
Sorry for that. I'll generate the samples.pkl file again for you, guys. I'll try to make it today.
You can download **samples.pkl** [here](https://we.tl/t-4aQmNrpO9T). The dump has already the dataset divided in train/test. So, you can call like this: ```py x_train, x_test, y_train, y_test = pkl.load(open('data/sample.pkl', 'rb')) ```
I uploaded a new samples.pkl inside a .rar file available [here](https://drive.google.com/open?id=1lxwO-A-VBZCVuKKP1FgE_h0ZgfxUvlR4).
@sf308, are you normalizing (StandardScaler, MinMaxScaler) the training and test sets on the same way?
Can you run the notebook from beginning up to the end, @UmarZubair?
Please, take a look at my [Medium article](https://medium.com/analytics-vidhya/face-pose-estimation-with-deep-learning-eebd0e62dbaf), @sandhyacs