In [1]:

    

from scipy.io import loadmat


    

In [2]:

    

import numpy as np


    

In [77]:

    

import numpy.ma as ma


    

In [3]:

    

from numpy import linalg as LA


    

In [21]:

    

from PIL import Image


    

In [34]:

    

def show_image(npimg):
    return Image.fromarray(npimg.astype(np.uint8))


    

In [35]:

    

def show_normals(npnorms):
    return Image.fromarray(((npnorms+1)/2*255).astype(np.uint8))


    

In [4]:

    

mat = loadmat('../Code/Experiments/Outputs/VGG16.mat')


    

In [5]:

    

mat['Normals'].shape


    

    
Out[5]:





(654, 240, 320, 3)

In [6]:

    

mat['Predictions'].shape


    

    
Out[6]:





(654, 240, 320, 3)

In [7]:

    

shape = mat['Normals'].shape


    

In [8]:

    

Norms = mat['Normals']


    

In [9]:

    

Norms = np.divide(Norms,np.reshape(LA.norm(Norms,axis=3), (shape[0],shape[1],shape[2],1)))


    

    




/home/raman/Applications/miniconda3/lib/python3.6/site-packages/ipykernel_launcher.py:1: RuntimeWarning: invalid value encountered in true_divide
  """Entry point for launching an IPython kernel.

In [11]:

    

Norms = np.nan_to_num(Norms)


    

In [12]:

    

np.unique(LA.norm(Norms,axis=3))


    

    
Out[12]:





array([ 0.        ,  0.99999988,  0.99999994,  1.        ,  1.00000012], dtype=float32)

In [45]:

    

show_normals(Norms[39])


    

    
Out[45]:

In [80]:

    

Masks = np.all(Norms,axis=3)


    

In [133]:

    

show_image(Masks[39]*255)


    

    
Out[133]:

In [46]:

    

Preds = mat['Predictions']


    

In [47]:

    

Preds = np.divide(Preds,np.reshape(LA.norm(Preds,axis=3), (shape[0],shape[1],shape[2],1)))


    

    




/home/raman/Applications/miniconda3/lib/python3.6/site-packages/ipykernel_launcher.py:1: RuntimeWarning: invalid value encountered in true_divide
  """Entry point for launching an IPython kernel.

In [48]:

    

Preds = np.nan_to_num(Preds)


    

In [49]:

    

np.unique(LA.norm(Preds,axis=3))


    

    
Out[49]:





array([ 0.        ,  0.99999988,  0.99999994,  1.        ,  1.00000012], dtype=float32)

In [51]:

    

show_normals(Preds[39])


    

    
Out[51]:

In [52]:

    

Dot = np.sum(np.multiply(Norms,Preds),axis=-1)


    

In [62]:

    

Dot.shape


    

    
Out[62]:





(654, 240, 320)

In [53]:

    

Dot = np.clip(Dot, -1,1)


    

In [63]:

    

show_normals(Dot[39])


    

    
Out[63]:

In [136]:

    

show_normals(Preds[39]*np.expand_dims(Masks[39], axis=2))


    

    
Out[136]:

In [137]:

    

show_normals(Dot[39]*Masks[39])


    

    
Out[137]:

In [138]:

    

Err = ma.masked_array(np.rad2deg(np.arccos(Dot)), ~Masks)


    

In [139]:

    

Err.mean()


    

    
Out[139]:





41.355931809061389

In [141]:

    

np.mean(np.less(Err, 11.25).astype(np.float32))*100


    

    
Out[141]:





12.458500597629401

In [142]:

    

Err = Err.compressed()


    

In [143]:

    

Summary = {'Mean':float(np.mean(Err)),
          'Median':float(np.median(Err)),
          'RMSE':float(np.sqrt(np.mean(np.power(Err,2)))),
          '11.25':np.mean(np.less(Err, 11.25).astype(np.float32))*100,
          '22.5':np.mean(np.less(Err, 22.5).astype(np.float32))*100,
          '30':np.mean(np.less(Err, 30).astype(np.float32))*100,
          '45':np.mean(np.less(Err, 45).astype(np.float32))*100}


    

In [144]:

    

Summary


    

    
Out[144]:





{'11.25': 12.458500266075134,
 '22.5': 30.244261026382446,
 '30': 41.094136238098145,
 '45': 59.958332777023315,
 'Mean': 41.35591125488281,
 'Median': 36.82966995239258,
 'RMSE': 49.13106918334961}

In [145]:

    

import json


    

In [146]:

    

json.dumps(Summary)


    

    
Out[146]:





'{"Mean": 41.35591125488281, "Median": 36.82966995239258, "RMSE": 49.13106918334961, "11.25": 12.458500266075134, "22.5": 30.244261026382446, "30": 41.094136238098145, "45": 59.958332777023315}'

In [147]:

    

%%writefile ../Code/Experiments/Evaluation.py
# Imports
import json
import numpy as np
import numpy.ma as ma
from numpy import linalg as LA

def Evaluate(ID, groundTruth, predictions):
    shape = groundTruth.shape
    # Normalization
    Norms = np.divide(groundTruth,np.reshape(LA.norm(groundTruth,axis=3), (shape[0],shape[1],shape[2],1)))
    Preds = np.divide(predictions,np.reshape(LA.norm(predictions,axis=3), (shape[0],shape[1],shape[2],1)))
    Norms = np.nan_to_num(Norms)
    Preds = np.nan_to_num(Preds)
    # Mask of valid values
    Masks = np.all(Norms,axis=3)
    # Dot product
    Dot = np.sum(np.multiply(Norms,Preds),axis=-1)
    Dot = np.clip(Dot, -1,1)
    # Error
    Err = np.rad2deg(np.arccos(Dot))
    Err = ma.masked_array(Err, ~Masks)
    Err = Err.compressed()
    # Stats
    Stats = {
        'Mean':float(np.mean(Err)),
        'Median':float(np.median(Err)),
        'RMSE':float(np.sqrt(np.mean(np.power(Err,2)))),
        '11.25':np.mean(np.less(Err, 11.25).astype(np.float32))*100,
        '22.5':np.mean(np.less(Err, 22.5).astype(np.float32))*100,
        '30':np.mean(np.less(Err, 30).astype(np.float32))*100,
        '45':np.mean(np.less(Err, 45).astype(np.float32))*100,}
    # Saving the results
    with open('Experiments/Outputs/'+ ID + '.eval', 'w') as fp:
        json.dump(Stats, fp)


    

    




Overwriting ../Code/Experiments/Evaluation.py

In [ ]: