A data table for all trials run in the first study
| Input frame rate (Hz) | Mean output frame rate (Hz) | SD output frame rate (Hz) | Mean delay (ms) | SD delay (ms) | Mean accuracy | SD accuracy | n |
|---|---|---|---|---|---|---|---|
| 90 | 46.95 | 1.53 | 34.34 | 5.94 | 0.97 | 0.01 | 13 |
| 180 | 68.13 | 5.10 | 39.85 | 42.97 | 0.94 | 0.02 | 13 |
| 200 | 70.17 | 7.07 | 32.56 | 10.70 | 0.95 | 0.00 | 13 |
| 220 | 69.31 | 0.33 | 33.82 | 15.10 | 0.95 | 0.03 | 6 |
| 270 | 80.83 | 18.27 | 55.93 | 55.69 | 0.94 | 0.01 | 13 |
| 300 | 91.45 | 5.27 | 60.62 | 77.71 | 0.94 | 0.01 | 11 |
| 320 | 109.76 | 49.24 | 53.28 | 43.80 | 0.93 | 0.04 | 9 |
Input frame rate, the streaming frame rate set in the software for our camera, representing the frame rate of the camera without any additional analyses; mean output frame rate, the mean number of frames per second processed by our system across all recordings at each input frame rate; SD output frame rate, the SD in the number of frames per second processed by our system across all recordings at each input frame rate; mean delay, the mean amount of time, in milliseconds, between the system receiving a frame that contains a left paw movement exceeding 5 pixels of vertical movement and the system sending a signal to the breakout board to trigger the LED to provide feedback for that movement; SD delay, the SD, in milliseconds, of the delay discussed above; mean accuracy, the mean output of the sigmoid function by TensorFlow for all body parts for all trials at each frame rate; SD accuracy, the SD of the output of the aforementioned sigmoid function; n, the number of trials recorded at each frame rate.