17 KiB
17 KiB
In [1]:
## USE for Multi GPU Systems #import os #os.environ["CUDA_VISIBLE_DEVICES"]="0" import numpy as np import matplotlib.pyplot as plt import pandas as pd import math import tensorflow as tf %matplotlib inline # Importing SK-learn to calculate precision and recall import sklearn from sklearn import metrics target_names = ["Knuckle", "Finger"]
Using TensorFlow backend.
In [2]:
df = pd.read_pickle("DataStudyCollection/df_statistics.pkl") lst = df.userID.unique() np.random.seed(42) np.random.shuffle(lst) test_ids = lst[-5:] train_ids = lst[:-5] print(train_ids, test_ids) df["GestureId"] = df.TaskID % 17 #df_train = dfAll[dfAll.userID.isin(train_ids)] #df_test = dfAll[dfAll.userID.isin(test_ids)] x = np.concatenate(df.Blobs.values).reshape(-1,27,15,1) x = x / 255.0 # convert class vectors to binary class matrices (one-hot notation) num_classes = len(df.TaskID.unique()) y = tf.keras.utils.to_categorical(df.TaskID, num_classes) labels = sorted(df.TaskID.unique())
[ 1 2 9 6 4 14 17 16 12 3 10 18 5] [13 8 11 15 7]
In [3]:
x = np.stack(df.Blobs) x = x.reshape(-1, 27, 15, 1)
In [4]:
# convert class vectors to binary class matrices (one-hot notation) num_classes = 2 y = tf.keras.utils.to_categorical(df.InputMethod, num_classes)
In [5]:
i = 1 plt.imshow(x[i].reshape(27, 15)) #np.sqrt(784) = 28 plt.title("Label for image %i is: %s" % (i, y[i]))
Out[5]:
Text(0.5, 1.0, 'Label for image 1 is: [1. 0.]')
In [6]:
# If GPU is not available: # GPU_USE = '/cpu:0' #config = tf.ConfigProto(device_count = {"GPU": 1}) # If GPU is available: config = tf.ConfigProto() config.log_device_placement = True config.allow_soft_placement = True config.gpu_options.allow_growth=True config.gpu_options.allocator_type = 'BFC' # Limit the maximum memory used config.gpu_options.per_process_gpu_memory_fraction = 0.4 # set session config tf.keras.backend.set_session(tf.Session(config=config))
In [7]:
loadpath = "./ModelSnapshots/CNN-33767.h5" model = tf.keras.models.load_model(loadpath)
In [8]:
%%time lst = [] batch = 100 for i in range(0, len(x), batch): _x = x[i: i+batch] lst.extend(model.predict(_x))
CPU times: user 1min 45s, sys: 12 s, total: 1min 57s Wall time: 1min 12s
In [9]:
df["InputMethodPred"] = lst df.InputMethodPred = df.InputMethodPred.apply(lambda x: np.argmax(x))
In [10]:
df = df.groupby(["userID", "TaskID", "VersionID"])[["InputMethodPred", "InputMethod"]].agg(lambda x: x.tolist()).reset_index()
In [11]:
from collections import Counter
In [12]:
df.InputMethod = df.InputMethod.apply(lambda x: Counter(x).most_common()[0][0]) df.InputMethodPred = df.InputMethodPred.apply(lambda x: Counter(x).most_common()[0][0])
In [13]:
df.groupby(["InputMethod", "InputMethodPred"]).count()
Out[13]:
| userID | TaskID | VersionID | ||
|---|---|---|---|---|
| InputMethod | InputMethodPred | |||
| 0 | 0 | 4425 | 4425 | 4425 |
| 1 | 124 | 124 | 124 | |
| 1 | 0 | 83 | 83 | 83 |
| 1 | 5935 | 5935 | 5935 |
In [14]:
df_train = df[df.userID.isin(train_ids)] df_test = df[df.userID.isin(test_ids)]
In [15]:
print(sklearn.metrics.confusion_matrix(df_train.InputMethod.values, df_train.InputMethodPred.values, labels=[0, 1])) cm = sklearn.metrics.confusion_matrix(df_train.InputMethod.values, df_train.InputMethodPred.values, labels=[0, 1], ) cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis] print(cm) print("Accuray: %.3f" % sklearn.metrics.accuracy_score(df_train.InputMethod.values, df_train.InputMethodPred.values)) print("Recall: %.3f" % metrics.recall_score(df_train.InputMethod.values, df_train.InputMethodPred.values)) print("Precision: %.3f" % metrics.average_precision_score(df_train.InputMethod.values, df_train.InputMethodPred.values)) print("F1-Score: %.3f" % metrics.f1_score(df_train.InputMethod.values, df_train.InputMethodPred.values)) print(sklearn.metrics.classification_report(df_train.InputMethod.values, df_train.InputMethodPred.values, target_names=target_names))
[[3260 82]
[ 47 4274]]
[[0.97546379 0.02453621]
[0.01087711 0.98912289]]
Accuray: 0.983
Recall: 0.989
Precision: 0.977
F1-Score: 0.985
precision recall f1-score support
Knuckle 0.99 0.98 0.98 3342
Finger 0.98 0.99 0.99 4321
micro avg 0.98 0.98 0.98 7663
macro avg 0.98 0.98 0.98 7663
weighted avg 0.98 0.98 0.98 7663
In [16]:
print(sklearn.metrics.confusion_matrix(df_test.InputMethod.values, df_test.InputMethodPred.values, labels=[0, 1])) cm = sklearn.metrics.confusion_matrix(df_test.InputMethod.values, df_test.InputMethodPred.values, labels=[0, 1], ) cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis] print(cm) print("Accuray: %.3f" % sklearn.metrics.accuracy_score(df_test.InputMethod.values, df_test.InputMethodPred.values)) print("Recall: %.3f" % metrics.recall_score(df_test.InputMethod.values, df_test.InputMethodPred.values)) print("Precision: %.3f" % metrics.average_precision_score(df_test.InputMethod.values, df_test.InputMethodPred.values)) print("F1-Score: %.3f" % metrics.f1_score(df_test.InputMethod.values, df_test.InputMethodPred.values)) print(sklearn.metrics.classification_report(df_test.InputMethod.values, df_test.InputMethodPred.values, target_names=target_names))
[[1165 42]
[ 36 1661]]
[[0.96520298 0.03479702]
[0.02121391 0.97878609]]
Accuray: 0.973
Recall: 0.979
Precision: 0.967
F1-Score: 0.977
precision recall f1-score support
Knuckle 0.97 0.97 0.97 1207
Finger 0.98 0.98 0.98 1697
micro avg 0.97 0.97 0.97 2904
macro avg 0.97 0.97 0.97 2904
weighted avg 0.97 0.97 0.97 2904