OBJECT DETECTION APPARATUS AND METHOD

• US 20140093129A1
• Filed: 09/27/2013
• Published: 04/03/2014
• Est. Priority Date: 10/01/2012
• Status: Active Grant

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation), Toshiba Digital Solutions Corporation (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: OKUMURA, Mayu, WATANABE, Tomoki, SEKI, Akihito
• Granted Patent: US 9,489,737 B2
• Time in Patent Office: Days
• US Class Current: 382/103
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/30201 Face
• CPC Class Codes: G06T 7/215 Motion-based segmentation
• CPC Class Codes: G06T 7/248 involving reference images or patches
• CPC Class Codes: G06T 7/32 using correlation-based methods
• CPC Class Codes: G06T 7/33 using feature-based methods
• CPC Class Codes: G06T 7/70 Determining position or orientation of objects or cameras camera calibration G06T7/80
• CPC Class Codes: G06T 7/97 Determining parameters from multiple pictures depth or shape recovery from multiple images G06T7/55; stereo camera calibration G06T7/85

Abstract

According to one embodiment, an object detection apparatus includes an acquisition unit, a first detector, a determination unit, and a second detector. The acquisition unit acquires frames in a time-series manner. The first detector detects a predetermined object in each of the frames. The determination unit stores detection results corresponding to the frames, compares a first detection result corresponding to a first frame of the frames with a second detection result corresponding to a second frame of the frames, and determines whether false negative of the predetermined object exists in the second frame. The second detector detects the predetermined object in the second frames when it is determined that false negative of the predetermined object exists. The second detector differs in performance from the first detector.

Claims

1. An object detection apparatus comprising:

an acquisition unit configured to acquire a plurality of frames in a time-series manner;

a first detector configured to detect a predetermined object in each of the frames;

a determination unit configured to store detection results corresponding to the frames, to compare a first detection result corresponding to a first frame of the frames with a second detection result corresponding to a second frame of the frames acquired after the first frame, and to determine whether false negative of the predetermined object exists in the second frame; and

a second detector configured to detect the predetermined object in the second frame when it is determined that false negative of the predetermined object exists, the second detector differing in performance from the first detector.

2. The apparatus according to claim 1, further comprising a storage configured to store a detection result of the first detector and a detection result of the second detector,

wherein the determination unit compares the detection result of the second detector corresponding to the second frame, with the detection result of the first detector corresponding to a third frame acquired after the second frame, and determine whether false negative of the predetermined object exists in the third frame.

3. The apparatus according to claim 1, wherein the determination unit employs a current frame as the second frame, employs, as the first frame, a previous frame before the current frame, and determines that false negative exists in the current frame, when a first number of objects detected in the current frame is larger than a second number of objects detected in the previous frame.

4. The apparatus according to claim 1, wherein the determination unit employs a current frame as the second frame, employs, as the first frame, a previous frame before the current frame, and determines that false negative of the predetermined object exists in the current frame, when a distance between a position of the predetermined object in the current frame, and a position of the predetermined object in the previous frame is larger than a threshold.

5. The apparatus according to claim 1, wherein the determination unit employs a current frame as the second frame, employs, as the first frame, a previous frame before the current frame, and determines that false negative of the predetermined object exists in the current frame, when a ratio of a fourth area to a third area is smaller than a threshold, the third area being a sum set of a first area including the predetermined object detected in the previous frame and a second area including the predetermined object detected in the current frame, the fourth area being a product set of the first area and the second area.

6. The apparatus according to claim 1, wherein the determination unit employs a current frame as the second frame, employs, as the first frame, a previous frame before the current frame, and determines that false negative of the predetermined object exists in the current frame, when a degree of similarity between a first color of the predetermined object detected in the current frame and a second color of the predetermined object detected in the previous frame, is smaller than a threshold.

7. The apparatus according to claim 1, further comprising a learning unit configured to update a parameter used for detection, using feature calculated from a first area including the predetermined object in the second frame, and a second area including part of the predetermined object or excluding the predetermined object,

wherein the second detector detects, using the parameter, whether each of the frames contains the predetermined object.

8. The apparatus according to claim 7, wherein the first area is detected by at least one of the first detector and the second detector, and the second area is obtained by randomly moving the first area.

9. The apparatus according to claim 7, wherein the learning unit updates the parameter by online learning.

10. The apparatus according to claim 2, wherein the determination unit employs a current frame as the second frame, employs, as the first frame, a previous frame before the current frame, and determines that false negative exists in the current frame, when a first number of objects detected in the current frame is larger than a second number of objects detected in the previous frame.

11. The apparatus according to claim 2, wherein the determination unit employs a current frame as the second frame, employs, as the first frame, a previous frame before the current frame, and determines that false negative of the predetermined object exists in the current frame, when a distance between a position of the predetermined object in the current frame, and a position of the predetermined object in the previous frame is larger than a threshold.

12. The apparatus according to claim 2, wherein the determination unit employs a current frame as the second frame, employs, as the first frame, a previous frame before the current frame, and determines that false negative of the predetermined object exists in the current frame, when a ratio of a fourth area to a third area is smaller than a threshold, the third area being a sum set of a first area including the predetermined object detected in the previous frame and a second area including the predetermined object detected in the current frame, the fourth area being a product set of the first area and the second area.

13. The apparatus according to claim 2, wherein the determination unit employs a current frame as the second frame, employs, as the first frame, a previous frame before the current frame, and determines that false negative of the predetermined object exists in the current frame, when a degree of similarity between a first color of the predetermined object detected in the current frame and a second color of the predetermined object detected in the previous frame, is smaller than a threshold.

14. An object detection method comprising

acquiring a plurality of frames in a time-series manner;

detecting a predetermined object in each of the frames, using a detector;

storing detection results corresponding to the frames, comparing a detection result corresponding to a first frame of the frames with a detection result corresponding to a second frame of the frames acquired after the first frame, and determining whether false negative of the predetermined object exists in the second frame; and

detecting the predetermined object in the second frame using another detector differing in performance from the first-mentioned detector, when it is determined that false negative of the predetermined object exists.