.. code:: ipython3

    import sys
    if '..' not in sys.path:
        sys.path.append('..')

.. code:: ipython3

    import tensorflow as tf
    from ultrayolo import YoloV3, datasets
    from ultrayolo.helpers import draw
    from pathlib import Path
    import numpy as np
    import logging
    
    from matplotlib import patches
    import matplotlib.pyplot as plt

Predict using a custom Model
============================

The classes contained in the dataset

.. code:: ipython3

    classes_dict = datasets.load_classes('./custom_classes.txt', True)
    target_shape = (512, 512, 3)
    max_objects = 100
    num_classes = len(classes_dict)
    print(f'number of classes {num_classes}')
    classes_dict


.. parsed-literal::

    number of classes 5




.. parsed-literal::

    {0: 'book',
     1: 'bus',
     2: 'car',
     3: 'motorcycle',
     4: 'vehicle registration plate'}



.. code:: ipython3

    model = YoloV3(target_shape, max_objects,
                   num_classes=num_classes, score_threshold=0.7, iou_threshold=0.7, 
                   training=False, backbone='DarkNet')

.. code:: ipython3

    tf.keras.utils.plot_model(model.model, show_shapes=True)




.. image:: 2_predict_custom_model_files/2_predict_custom_model_5_0.png



Load the weights
----------------

load a custom model from `here <add%20a%20valid%20link%20here>`__

.. code:: ipython3

    w_path = Path('./weights.117-9.932.h5')
    model.model.load_weights(str(w_path))
    # model.load_weights(w_path)

Predict
-------

we predict the objects using an image from the web. You can try with
your.

Download an image
~~~~~~~~~~~~~~~~~

.. code:: ipython3

    img = datasets.open_image('https://c8.staticflickr.com/4/3901/14855908765_8bdda9130b_z.jpg')
    # img = datasets.open_image('https://lh6.googleusercontent.com/proxy/Jo961aR6HemjY-D0TKiVEkVlI7b84uTkfJHSFBCz4UN2maJidjYVznbPrxDpRDd6wlbqn80ZmP_ohdCPkE9syrVJPIjiYvgbo9ovRAArlFC_9Sm4V3NZi--R')
    img_pad = datasets.pad_to_fixed_size(img, target_shape)
    img_resized = datasets.resize(img, target_shape)
    #preprocess the image
    x = np.divide(img_pad, 255.)
    x = np.expand_dims(x, 0)
    x.shape




.. parsed-literal::

    (1, 512, 512, 3)



Perform the prediction
~~~~~~~~~~~~~~~~~~~~~~

.. code:: ipython3

    boxes, scores, classes, sel = model.predict(x)
    print(f'found {sel[0]} objects')


.. parsed-literal::

    found 3 objects


Uncomment the cells below to see what the model returns

boxes[:,:sel[0],:]

scores

classes

Show the image with the discovered objects
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

.. code:: ipython3

    ax = draw.show_img(img_resized, figsize=(16,10))
    for i,b in enumerate(boxes[0,:sel[0]]):
        draw.rect(ax, b, color='#9cff1d')
        name_score = f'{classes_dict[classes[0, i]]} {str(round(scores[0,i],2))}'
        draw.text(ax, b[:2], name_score, sz=12)
        print(classes_dict[classes[0, i]], scores[0,i])
        
    plt.show()


.. parsed-literal::

    car 0.97570795
    car 0.9543877
    car 0.83372337



.. image:: 2_predict_custom_model_files/2_predict_custom_model_17_1.png