Method of and system for determining traffic signal state

A method for determining a predicted state of a traffic signal, the method executable on a server, the method comprising:

• receiving, by the server, a video of a plurality of traffic signals, wherein the plurality of traffic signals comprises the traffic signal;
• generating, based on the video, a plurality of still images of the plurality of traffic signals;
• determining, for each still image of the plurality of still images, and by a first Machine Learning Algorithm (MLA), a vector for each bulb in the respective still image, thereby generating a plurality of vectors corresponding to the video,
• each vector indicating a predicted likelihood of each of a plurality of statuses of the respective bulb,
• the first MLA having been trained based on a plurality of labeled still images of traffic signals, and
• each labeled still image of the plurality of labeled still images indicating a status of each bulb of the respective labeled still image;
• determining, based on the plurality of vectors, and by a second MLA, a predicted state of each of the plurality of traffic signals,
• the second MLA having been trained based on a plurality of labeled vectors, wherein each labeled vector indicating a predicted likelihood of each of the plurality of statuses for a given bulb of a given traffic signal, and
• wherein each labeled vector is associated with a label indicating an actual state of the given traffic signal;
• determining, based on the predicted state of each of the plurality of traffic signals, the predicted state of the traffic signal;
• and
• storing an indicator of the predicted state of the traffic signal.

A system for determining a predicted state of a traffic signal, the system comprising:

• a processor;
• and
• a non-transitory computer-readable medium comprising instructions,
• the processor, upon executing the instructions, being configured to:
• receive a video of the traffic signal;
• generate, based on the video, a plurality of still images of the traffic signal;
• determine, for each still image of the plurality of still images, and by a first Machine Learning Algorithm (MLA), a vector for each bulb in the respective still image, thereby generating a plurality of vectors corresponding to the video,
• wherein each vector indicates a predicted likelihood of each of a plurality of statuses of the respective bulb,
• wherein the first MLA is trained based on a plurality of labeled still images of traffic signals, and
• wherein each labeled still image of the plurality of labeled still images indicates a status of each bulb of the respective labeled still image;
• determine, based on the plurality of vectors, and by a second MLA, the predicted state of the traffic signal,
• wherein the second MLA is trained based on a plurality of labeled vectors,
• wherein each labeled vector indicates a predicted likelihood of each of the plurality of statuses for a bulb of a traffic signal, and
• wherein each labeled vector is associated with a label indicating a state of the traffic signal;
• and
• store an indicator of the predicted state of the traffic signal.