Apparatus, method, and computer program for identifying state of lighting

A method for identifying the state of lighting, comprising:

• identifying the states of lighting of a first lighting part and a second lighting part of a vehicle that is moving by inputting time-series images representing the vehicle that is moving into a classifier comprising a feature calculation part, a first state-of-lighting identification part including a recursive structure, and a second state-of-lighting identification part including a recursive structure, wherein
• the feature calculation part of the classifier calculates, whenever one of the time-series images is inputted in chronological order, a feature map representing features of the first lighting part and the second lighting part by executing a convolution operation on the inputted image,
• the first state-of-lighting identification part of the classifier identifies the state of lighting of the first lighting part while updating a first internal state that is recursively used by inputting the feature map calculated for each of the time-series images in chronological order,
• the second state-of-lighting identification part of the classifier identifies the state of lighting of the second lighting part while updating a second internal state that is recursively used by inputting the feature map calculated for each of the time-series images in chronological order, wherein the first lighting part is a direction indicator, and the second lighting part is a rotating light, and
• the method further comprises:
• detecting an object region representing the vehicle that is moving from each of the time-series images;
• tracking the vehicle that is moving to associate object regions representing the vehicle that is moving in the respective time-series images with each other;
• and
• inputting the object regions representing the vehicle that is moving in the respective time-series images into the classifier.

A non-transitory recording medium that stores a computer program for identifying the state of lighting, the computer program causing a computer to execute a process comprising:

• identifying the states of lighting of a first lighting part and a second lighting part of a vehicle that is moving by inputting time-series images representing the vehicle that is moving into a classifier comprising a feature calculation part, a first state-of-lighting identification part including a recursive structure, and a second state-of-lighting identification part including a recursive structure, wherein
• the feature calculation part of the classifier calculates, whenever one of the time-series images is inputted in chronological order, a feature map representing features of the first lighting part and the second lighting part by executing a convolution operation on the inputted image,
• the first state-of-lighting identification part of the classifier identifies the state of lighting of the first lighting part while updating a first internal state that is recursively used by inputting the feature map calculated for each of the time-series images in chronological order, and
• the second state-of-lighting identification part of the classifier identifies the state of lighting of the second lighting part while updating a second internal state that is recursively used by inputting the feature map calculated for each of the time-series images in chronological order, wherein the first lighting part is a direction indicator, and the second lighting part is a rotating light, and
• the computer program causes the computer to further execute:
• detecting an object region representing the vehicle that is moving from each of the time-series images;
• tracking the vehicle that is moving to associate object regions representing the vehicle that is moving in the respective time-series images with each other;
• and
• inputting the object regions representing the vehicle that is moving in the respective time-series images into the classifier.