Granted Patent B2

US 12,367,594 · App. 18/070,653 · Granted Jul 22, 2025

Inventors: Manabu Nagai (Nagoya, JP); Suguru Watanabe (Nagoya, JP); Yusuke Nakano (Nagoya, JP)

Assignee: TOYOTA JIDOSHA KABUSHIKI KAISHA

G06T7/20G06T7/70G06T2207/10016G06T2207/30196

Abstract

Claims (19)

1. An evaluation system configured to evaluate a quality of an algorithm that estimates the positions of movable objects in a video image and tracks the objects, the evaluation system comprising:

a processor configured to:

determine, for each of the objects, a correct estimation result or one of a plurality of false estimation types, which indicate types of false estimation results using ground truth data that corresponds to the video image and output data indicating the result of the estimation made on the video image by the algorithm; and

add false estimation coefficients that correspond to the plurality of respective false estimation types and are provided so as to become higher in accordance with a degree of impact of the false estimation type for a number of objects that correspond to the false estimation type and thus calculate an evaluation value of the algorithm based on a total value of the added values of the false estimation coefficients obtained for each of the plurality of false estimation types, and

wherein the false estimation coefficient that corresponds to a second false estimation, which is one of the plurality of false estimation types and is a false estimation type regarding a situation in which the algorithm estimates the plurality of objects included in the ground truth data between one timing and the next timing in a switched manner, is set to be higher than the false estimation coefficient that corresponds to a third false estimation, which is a false estimation type regarding a situation in which the algorithm estimates that one of the objects included in the ground truth data at one timing is different from the object included in the ground truth data at the next timing.

2. The evaluation system according to claim 1 , wherein the false estimation coefficient that corresponds to a first false estimation, which is one of the plurality of false estimation types and is a false estimation type regarding a situation in which the algorithm has failed to estimate the object included in the ground truth data, is set to be higher than the false estimation coefficients that correspond to the other false estimation types.

3. The evaluation system according to claim 1 , wherein the processor is configured to calculate the false estimation coefficient for each timing of estimation.

4. The evaluation system according to claim 3 , wherein the processor is configured to calculate when a false estimation of the same type has been continuously made for one object, the false estimation coefficient in such a way that the false estimation coefficient that corresponds to this false estimation type regarding this object is made higher.

5. The evaluation system according to claim 4 , wherein the processor is configured to calculate, when a first false estimation, which is a false estimation type regarding a situation in which the algorithm has failed to estimate the object included in the ground truth data, has been continuously made, the false estimation coefficient in such a way that the false estimation coefficient that corresponds to the first false estimation is made higher.

6. The evaluation system according to claim 3 , wherein the processor is configured to calculate the false estimation coefficient in such a way that the false estimation coefficient that corresponds to a false estimation type regarding the object becomes higher as the speed of this object increases.

7. The evaluation system according to claim 3 , wherein the processor is configured to calculate the false estimation coefficient in such a way that the false estimation coefficient that corresponds to a false estimation type regarding an object becomes lower as the number of other objects whose type is the same as that of the object and whose distances from this object are equal to or smaller than a predetermined threshold becomes larger.

8. An evaluation method for evaluating a quality of an algorithm that estimates the positions of movable objects in a video image and tracks the objects, the evaluation method comprising:

determining, for each of the objects, a correct estimation result or one of a plurality of false estimation types, which indicate types of false estimation results, using ground truth data that corresponds to the video image and output data indicating the result of the estimation made on the video image by the algorithm; and

adding false estimation coefficients that correspond to the plurality of respective false estimation types and are provided so as to become higher in accordance with a degree of impact of the false estimation type for a number of objects that correspond to the false estimation type and thus calculating an evaluation value of the algorithm based on a total value of the added values of the false estimation coefficients obtained for each of the plurality of false estimation types, and

wherein the false estimation coefficient that corresponds to a second false estimation, which is one of the plurality of false estimation types and is a false estimation type regarding a situation in which the algorithm estimates the plurality of objects included in the ground truth data between one timing and the next timing in a switched manner, is set to be higher than the false estimation coefficient that corresponds to a third false estimation, which is a false estimation type regarding a situation in which the algorithm estimates that one of the objects included in the ground truth data at one timing is different from the object included in the ground truth data at the next timing.

9. A non-transitory computer readable medium storing a program for implementing an evaluation method for evaluating a quality of an algorithm that estimates the positions of movable objects in a video image and tracks the objects, the program causing a computer to execute the processing of:

determining, for each of the objects, a correct estimation result or one of a plurality of false estimation types, which indicate types of false estimation results, using ground truth data that corresponds to the video image and output data indicating the result of the estimation made on the video image by the algorithm; and

adding false estimation coefficients that correspond to the plurality of respective false estimation types and are provided so as to become higher in accordance with a degree of impact of the false estimation type for a number of objects that correspond to the false estimation type and thus calculating an evaluation value of the algorithm based on a total value of the added values of the false estimation coefficients obtained for each of the plurality of false estimation types, and

wherein the false estimation coefficient that corresponds to a second false estimation, which is one of the plurality of false estimation types and is a false estimation type regarding a situation in which the algorithm estimates the plurality of objects included in the ground truth data between one timing and the next timing in a switched manner, is set to be higher than the false estimation coefficient that corresponds to a third false estimation, which is a false estimation type regarding a situation in which the algorithm estimates that one of the objects included in the ground truth data at one timing is different from the object included in the ground truth data at the next timing.

Assignments (1)

Priority Claims (1)

Continuity (1)

References Cited (10)