Method for training a system for automated detection of the stroboscopic effect
A stroboscopic device, comprising: a camera for acquiring video frames; a light source; and at least one hardware processor; and software that is configured to, when executed by the at least one hardware processor, generating synthetic video frames of an object, providing real video frames of the object and the synthetic video frames to a discriminator configured to detect the difference between the real video frames from the synthetic video frames, using feedback from the discriminator to generate further synthetic video frames, providing real video frames of the object and the further synthetic video frames to the discriminator and repeating until a convergence point is achieved where the discriminator can't reliably tell the real video frames from the synthetically generated video frames, generating a data set comprising synthetic video frames and real vide frames of the object, using the data set to train a model, acquire a sequence of video frames via the camera, for a first and second frame in the sequence of video frames, compute the dense optical flow field between the two frames, wherein the optical flow field contains a flow vector for each pixel in each of the first and second frames, indicating the motion of that pixel from the first frame to the second frame, calculate the average magnitude (Average Optical Flow or AOF) of all flow vectors for each pixel in the first and second frames, compare the AOF to a threshold, and store a result based on the comparison, repeat the process for all frames in the sequence of video frames, determine whether a stroboscopic effect has been achieved based on the store results, control the activation of the light source until the results indicate that the stroboscopic effect has been achieved, and obtain a further sequence of video frames after the stroboscopic effect is achieved, identify the object in the further sequence of video frames using the model, and automatically detect movement or vibration of the identified object using the model.
1. A stroboscopic device, comprising:
a camera for acquiring video frames;
a light source; and
at least one hardware processor; and
software that is configured to, when executed by the at least one hardware processor,
generating synthetic video frames of an object,
providing real video frames of the object and the synthetic video frames to a discriminator configured to detect the difference between the real video frames from the synthetic video frames,
using feedback from the discriminator to generate further synthetic video frames,
providing real video frames of the object and the further synthetic video frames to the discriminator and repeating until a convergence point is achieved where the discriminator can't reliably tell the real video frames from the synthetically generated video frames,
generating a data set comprising synthetic video frames and real vide frames of the object,
using the data set to train a model,
acquire a sequence of video frames via the camera,
for a first and second frame in the sequence of video frames, compute the dense optical flow field between the two frames, wherein the optical flow field contains a flow vector for each pixel in each of the first and second frames, indicating the motion of that pixel from the first frame to the second frame,
calculate the average magnitude (Average Optical Flow or AOF) of all flow vectors for each pixel in the first and second frames,
compare the AOF to a threshold, and store a result based on the comparison,
repeat the process for all frames in the sequence of video frames,
determine whether a stroboscopic effect has been achieved based on the store results,
control the activation of the light source until the results indicate that the stroboscopic effect has been achieved, and
obtain a further sequence of video frames after the stroboscopic effect is achieved,
identify the object in the further sequence of video frames using the model, and
automatically detect movement or vibration of the identified object using the model.
2. The device of claim 1 , wherein the results are saved as a of positive or negative, and it is determined that the stroboscopic effect is achieved when the positive results exceed a second threshold.
3. The device of claim 2 , wherein the second threshold is 95%.
4. The device of claim 1 , wherein the light source is an LED light source.
5. The device of claim 1 , wherein converting the RGB image into a grayscale image comprising assigning weights to each of Red, Green, and Blue aspects of the RGB image, wherein weights are chosen to match the human perception of brightness, generate a weighted sum weighted sum for the Red, Green, and Blue aspects, wherein the weighted sum represents a grayscale pixel value.
6. The device of claim 1 , wherein the Farneback algorithm is used to determine the optical flow field.
7. The device of claim 1 , wherein the software is further configured to, when executed by the at least one hardware processor, generate visualizations of the optical flow field.
8. The device claim 1 , wherein the software is further configured to, when executed by the at least one hardware processor, identify multiple objects in the further sequence of video frames, and automatically detect movement or vibration for each of the identified objects.
9. The device of claim 1 , wherein in the model comprises a Super-Resolution Generative Adversarial Networks (SRGANs).