Denoising examples

This video compares denoising DVS data with our proposed  algorithms, by comparison with BAF and ONF.

Please watch the entire video - it has several segments. Best viewed full screen.

The yellow crosses show the ROC point: X coordinate is FPR=False Positive Rate (noise incorrectly passed). Y coordinate is TPR = True Positive Rate (signal correctly passed). Better denoising means closer to top left corner. The AUC (area under the curve) is obtained by sweeping the discrimination threshold over a large range; see Benchmarking.