Table 1. Performance for all categories using the proposed method (ConvNet & LRBSF).
Best and worst performance of individual participant is also mentioned.
| Average of 26 participants | Individual participant | ||||
|---|---|---|---|---|---|
| Feature selection with two sample t-test | Accuracy | Sensitivity | Specificity | Best accuracy | Worst accuracy |
| Human vs Animal | 77.1% | 77.2% | 76.9% | 87.1% | 73.1% |
| Human vs Building | 79.1% | 83.7% | 75.4% | 86.2% | 72.2% |
| Human vs Natural Scenes | 80.1% | 79.07% | 80.7% | 86.3% | 73.8% |
| Human vs Fruit | 78.7% | 75.5% | 81.2% | 89.1% | 73.3% |
| Animal vs Building | 81.5% | 76.56% | 88.88% | 87.4% | 72.5% |
| Animal vs Natural Scenes | 83.1% | 88.6% | 76.6% | 85.9% | 73.6% |
| Animal vs Fruit | 77.4% | 74.5% | 81.6% | 83.8% | 73.4% |
| Building vs Natural Scenes | 79.4% | 83.9% | 75.6% | 86.2% | 75.7% |
| Building vs Fruit | 81.5% | 85.4% | 78.5% | 89.7% | 73.6% |
| Natural Scenes vs Fruit | 81.1% | 74.4% | 85.97% | 88.8% | 74.2% |