Table 9.
Comparison of existing datasets.
| Paper | Model | Quantity of Data | Types of Data Used | Feature | Metric | Result |
|---|---|---|---|---|---|---|
| [3] | SVM | Total duration of around 5 min | Chainsaw sounds with background noise | MFCC | Accuracy | 91.07% |
| [11] | Random forest | 40 | Bird sounds, mammal sounds, insect sounds from Freesound | Double features | Average accuracy rates in different environments (rain, wind, traffic, average) | 86.28% |
| [51] | Cyclic HMM | 1418 | Animal sounds from HU-ASA database | MFCC | Accuracy | 64% |
| [4] | Configuration based on a CNN | 280 | Chainsaw sounds, chirping birds, crackling fire, crickets, handsaw, rain, and wind extracted from ESC50 | MFCC | Accuracy | 85.37% |
| [6] | SVM with log kernel | 3265 | Chainsaw sounds | MFCC | TPR | 53.16% |
| [5] | Feed-forward network | 217 | Chainsaw sounds, vehicle/engine sounds, forest sounds, urban sounds | Fourier power spectrum coefficients | Accuracy | 79.50% |
| [31] | CNN | 100 | Chainsaw | Fourier spectrogram | Accuracy | 96% |
| This Study | CNN | 2025 | 27 unique classes | Mel spectrogram | Accuracy | 92.59% |