Data Science Asked by Abdelmalek Mallek on May 28, 2021
I’m begginer in deep learning so I tried to execute a code of liveness face detection from github in this link :https://github.com/imironica/liveness , so when I tried to run features extraction with LBP and HAralik, I got a multiple results in confusion matrixes like this:
1- test features with LBP:
Train Nearest neighbors (3)
Accuracy: 0.7621621621621621
Confusion matrix:
[[1625 45]
[ 703 772]]
Train SGD
Accuracy: 0.8158982511923688
Confusion matrix:
[[1588 82]
[ 497 978]]
Train Naive Bayes
Accuracy: 0.8756756756756757
Confusion matrix:
[[1544 126]
[ 265 1210]]
Train Decision Tree Classifier
Accuracy: 0.6791732909379968
Confusion matrix:
[[1569 101]
[ 908 567]]
Train Adaboost Classifier
Accuracy: 0.6515103338632751
Confusion matrix:
[[1630 40]
[1056 419]]
Train Gradient Boosting Classifier
Accuracy: 0.6419713831478537
Confusion matrix:
[[1622 48]
[1078 397]]
Train Random Forest Classifier
Accuracy: 0.7106518282988871
Confusion matrix:
[[1641 29]
[ 881 594]]
Train Extremelly Trees Classifier
Accuracy: 0.724006359300477
Confusion matrix:
[[1638 32]
[ 836 639]]
Train Linear SVM with C=1
Accuracy: 0.8076311605723371
Confusion matrix:
[[1593 77]
[ 528 947]]
Train SVM with C=10
Accuracy: 0.8082670906200318
Confusion matrix:
[[1591 79]
[ 524 951]]
2-and when I tried test features with Haralik I got:
Train Nearest neighbors (3)
Nearest neighbors (3): 0.753577106518283
[[1580 90]
[ 685 790]]
Train SGD
SGD: 0.7424483306836248
[[1598 72]
[ 738 737]]
Train Naive Bayes
Naive Bayes: 0.7093799682034976
[[1521 149]
[ 765 710]]
Train Decision Tree Classifier
Decision Tree Classifier : 0.6511923688394277
[[1527 143]
[ 954 521]]
Train Adaboost Classifier
Adaboost Classifier : 0.763751987281399
[[1537 133]
[ 610 865]]
Train Gradient Boosting Classifier
Gradient Boosting Classifier: 0.7939586645468999
[[1544 126]
[ 522 953]]
Train Random Forest Classifier
Random Forest Classifier: 0.7007949125596185
[[1564 106]
[ 835 640]]
Train Extremelly Trees Classifier
Extremelly Trees Classifier: 0.7364069952305247
[[1543 127]
[ 702 773]]
Train Linear SVM with C=0.01
Linear SVM with C=0.01 : 0.5303656597774244
Train Linear SVM with C=0.1
Linear SVM with C=0.1 : 0.729093799682035
Train Linear SVM with C=1
Linear SVM with C=1 : 0.7335453100158983
Train Linear SVM with C=10
Linear SVM with C=10 : 0.7869634340222575
Train Linear SVM with C=100
Linear SVM with C=100 : 0.7939586645468999
Train Linear SVM with C=500
Linear SVM with C=500 : 0.7936406995230525
Train Linear SVM with C=1000
Linear SVM with C=1000 : 0.7968203497615263
Train Linear SVM with C=2000
Linear SVM with C=2000 : 0.794912559618442
Train SVM with C=0.01
SVM with C=0.01 : 0.5310015898251192
Train SVM with C=0.1
SVM with C=0.1 : 0.5306836248012718
Train SVM with C=1
SVM with C=1 : 0.7297297297297297
Train SVM with C=10
SVM with C=10 : 0.7456279809220986
Train SVM with C=100
SVM with C=100 : 0.7802861685214626
Train SVM with C=500
SVM with C=500 : 0.7910969793322734
Train SVM with C=1000
SVM with C=1000 : 0.7926868044515103
Train SVM with C=2000
SVM with C=2000 : 0.7939586645468999
Train SVM with C=4000
SVM with C=4000 : 0.7926868044515103
Train SVM with C=10000
SVM with C=10000 : 0.7930047694753577
Train SVM with C=2000000
SVM with C=2000000 : 0.7364069952305247
so now I’m trying to understand what is the difference between LBP and Haralik in features extraction and which is the best to choose?
Get help from others!
Recent Answers
Recent Questions
© 2024 TransWikia.com. All rights reserved. Sites we Love: PCI Database, UKBizDB, Menu Kuliner, Sharing RPP