Table 3 Average accuracy (%) of different approaches based on different numbers of images and image categories.
From: A Parallel Adaboost-Backpropagation Neural Network for Massive Image Dataset Classification
Image categories | Classification approach | Number of images | |||||
|---|---|---|---|---|---|---|---|
500 | 1,000 | 2,000 | 5,000 | 10,000 | 20,000 | ||
5 | RF | 96.1 | 95.0 | 91.9 | 85.8 | 77.8 | 67.9 |
SVM | 96.3 | 95.1 | 92.2 | 86.5 | 78.6 | 69.3 | |
Adaboost-BP | 96.3 | 95.0 | 92.6 | 87.4 | 80.1 | 71.7 | |
The method of Shi et al. | 96.3 | 95.3 | 93.5 | 90.0 | 85.5 | 78.3 | |
Parallel BP | 100.0 | 99.6 | 99.4 | 99.2 | 97.5 | 94.6 | |
Parallel Adaboost-BP | 100.0 | 100.0 | 99.9 | 99.8 | 98.0 | 97.1 | |
10 | RF | 95.0 | 93.5 | 88.9 | 82.6 | 74.3 | 63.8 |
SVM | 95.1 | 93.9 | 89.7 | 83.8 | 76.1 | 65.2 | |
Adaboost-BP | 95.1 | 94.2 | 91.4 | 85.5 | 77.2 | 67.0 | |
The method of Shi et al. | 95.5 | 94.7 | 92.6 | 89.3 | 84.5 | 78.0 | |
Parallel BP | 99.7 | 99.5 | 99.3 | 99.1 | 96.8 | 94.0 | |
Parallel Adaboost-BP | 100.0 | 100.0 | 99.8 | 99.7 | 97.3 | 96.3 | |
20 | RF | 93.1 | 92.0 | 85.9 | 77.9 | 69.1 | 58.8 |
SVM | 93.5 | 92.0 | 86.7 | 79.3 | 70.4 | 60.5 | |
Adaboost-BP | 94.0 | 92.1 | 88.3 | 81.9 | 72.6 | 62.0 | |
The method of Shi et al. | 94.0 | 94.0 | 91.6 | 87.2 | 80.5 | 71.8 | |
Parallel BP | 99.5 | 99.4 | 99.0 | 97.3 | 94.1 | 90.8 | |
Parallel Adaboost-BP | 99.8 | 99.6 | 99.5 | 99.0 | 97.1 | 95.3 | |
50 | RF | 92.7 | 89.4 | 82.8 | 75.0 | 66.6 | 56.8 |
SVM | 92.8 | 90.1 | 83.9 | 76.9 | 68.1 | 58.3 | |
Adaboost-BP | 92.8 | 90.3 | 85.6 | 79.1 | 70.7 | 60.6 | |
The method of Shi et al. | 93.3 | 93.0 | 91.0 | 86.6 | 78.5 | 68.9 | |
Parallel BP | 98.8 | 98.7 | 98.0 | 97.0 | 93.8 | 89.6 | |
Parallel Adaboost-BP | 99.7 | 99.0 | 98.5 | 97.4 | 96.0 | 93.4 | |
100 | RF | 89.7 | 87.6 | 80.1 | 73.7 | 64.1 | 54.1 |
SVM | 90.3 | 88.0 | 81.9 | 74.2 | 65.8 | 56.0 | |
Adaboost-BP | 90.9 | 88.4 | 83.7 | 77.5 | 68.4 | 58.3 | |
The method of Shi et al. | 92.3 | 92.1 | 89.0 | 83.2 | 69.7 | 59.4 | |
Parallel BP | 98.9 | 97.6 | 95.0 | 93.9 | 90.2 | 83.1 | |
Parallel Adaboost-BP | 99.5 | 99.0 | 98.1 | 94.8 | 92.3 | 90.0 | |
200 | RF | 87.0 | 83.6 | 76.9 | 68.8 | 58.1 | 46.0 |
SVM | 87.1 | 84.3 | 77.8 | 70.0 | 60.2 | 49.2 | |
Adaboost-BP | 87.2 | 85.1 | 79.5 | 72.3 | 62.4 | 51.0 | |
The method of Shi et al. | 90.3 | 89.5 | 87.6 | 81.2 | 68.7 | 58.3 | |
Parallel BP | 98.5 | 97.1 | 93.9 | 90.7 | 85.3 | 78.2 | |
Parallel Adaboost-BP | 99.5 | 98.5 | 98.1 | 93.4 | 90.7 | 86.1 | |
