Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Status of Claims
Claims 1 – 16 remain pending.
Claim 2 is Amended
Claims 11 – 16 are new claims
Response to Amendment
The amendment filed 10/30/2025 overcomes the following objections/rejections.
112 (b) Rejection
Response to Arguments
Applicant's arguments filed October 30, 2025 with respect to claims 1 – 16
have been considered but are moot because the new grounds of rejection necessitated by applicant’s amendment.
Response to Arguments
Applicant's arguments filed October 30, 2025 with respect to claims 1 – 16 have been fully considered but they are not persuasive.
Response to Remarks
Applicant argues that Peng is silent on the following limitations below. Examiner respectfully disagrees for the reasons provided below:
In the Remarks (p. 12) regarding claim 1 and 9, applicants assert, “Peng fails to disclose or suggest filtering the tracklets that have been associated together on the basis of time distance or space distance between the tracklets. Further, Peng fails to disclose or suggest filtering the tracklets based on a similarity between the tracklets. As Peng generates tracking information based on the clustering result (preliminary association result) alone, the increased accuracy arising from the filtering recited in the independent claims is not realized in the process in Peng. Barburescu is relied on to teach extracting tracklets, but also fails to disclose or suggest the filtering noted above”.
Examiner respectfully disagrees because Peng in [Section – 2.3] discloses about similarity based filtering, “the Re-ID features of the selected detections are input into a classifier to get the similarity score between tracklets and detections. In [61], spatial and temporal attention mechanism are adopted in feature extraction, which make the network focus on the matching patterns of the input image pair”.
For the reasons above, the rejections of claims 1 – 10 as established in the last Office
Action (Non-Final, 07/30/2025) are proper and are hereby maintained and incorporated in this Office Action.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1, 4, 5, 9 and 10 are rejected under 35 U.S.C 103 as being unpatentable over Peng “State-Aware Re-Identification Feature for Multi-Target Multi-Camera Tracking" (hereinafter Peng) in view of Barburescu US Patent Application Publication No. US-20230245460-A1 (hereinafter Barburescu).
Regarding claim 1, 4 and 5, the same grounds of rejection based on Peng and Barburescu from the last Office Action (Non-Final, 07/30/2025) are applied here.
Regarding claim 9, apparatus claim 9 corresponds to method claim 1. Therefore, the rejection analysis and motivation to combine of claim 1 is applicable to claim 9 as set forth in the last Office Action (Non-Final, 07/30/2025).
Regarding claim 10, claim 10 is a computer-readable storage medium claim which
correspond to claim 1. Therefore, claim 10 is rejected for the same reason provided above for
claim 1 as set forth in the last Office Action (Non-Final, 07/30/2025).
Claims 2, 3, 11, 12, 13 and 15 are rejected under 35 U.S.C 103 as being unpatentable over Peng in view of Barburescu and further in view of Jeon Patent Application Publication No. KR-20180086716-A (hereinafter Jeon) and Johann Patent Application Publication No. GB-2557316-A (hereinafter Johann).
Regarding claim 2, Peng in the combination discloses the method according to claim 1, wherein the method filtering the tracklets in the set further comprises: a) sequentially determining whether to add each tracklet in the tracklet sequence into a candidate set (Peng in [Section – 3.3.4, Paragraph – 7] discloses, “Tracklet clustering aims to associate all the tracklets except the Disappeared ones ... At first, distance matrix MT −cluster is constructed, then a greedy algorithm is adopted to associate the tracklets with the distance threshold
θ
c
l
u
s
t
e
r
”).
The combination of Peng and Barburescu doesn’t disclose the following limitations as further recited in the claim.
Jeon discloses b) removing the earliest tracklet in the tracklet sequence to truncate the tracklet sequence (Jeon in [0032] discloses, “it is possible to remove a tracklet which appears only in frames less than a predetermined number”); d) further truncating the truncated tracklet sequence (Jeon in [0049] discloses, “leaving only the tracklet that appear in all nine frames, and the tracklet that appear only in less than nine frames”); e) iteratively performing steps c) and d), and performing step f) when a value obtained by subtracting 1 from the number of tracklets in the current truncated tracklet sequence in the step c) is less than or equal to the maximum number of tracklets in respective candidate sets have been obtained (Jeon in [0030] further discloses about truncating the number of tracklets, “a step of eliminating erroneously detected tracklet may be performed. In other words, to reduce the pool of objects to determine if the tracking of the object succeeds or fails, it eliminates the badly detected, unreliable, traffic. A non-stable tracklet refers to a tracklet having a length less than a predetermined threshold”); and f) taking the candidate set including the maximum number of tracklets among the respective candidate sets that have been obtained in the step c) as the filtered set (Jeon in [0054] discloses final step of selecting candidate sets from filtered set, ““is possible to associate a dead tracklet with ID1 with a live tracklet with ID3, and reassign the ID of the associated object. That is, it reallocates ID3 to ID1”).
It would have been obvious to one with one having an ordinary skill in art before the effective filling date of the claimed invention to integrate the technique of Jeon into the system of Peng in view of Barburescu because it would allow the system to perform the function more efficiently avoiding unnecessary comparison.
The combination of Peng, Barburescu and Jeon doesn’t disclose the following limitations as further recited in the claim.
Johann discloses adding into the candidate set the tracklets that have been determined to be added into the candidate set (Johann in [Page – 23, Paragraph – 1] discloses, “In a step 1320, for each tracklet a set of candidates for merging is selected”. Furthermore, Johann in [Page – 24, Paragraph – 3] discloses, “Tracklet having a candidate with the lowest matching criterion are merged first. Then a merged candidate is removed from any other set of candidates it may belong to. In other words, the step of merging tracklets with its best candidate is done sequentially in increasing order of the value of the comparison score between each tracklet and its best candidate”), c) sequentially determining whether to add each tracklet in the truncated tracklet sequence into another candidate set , and adding into the other candidate set the tracklets that have been determined to be added into the other candidate set (Johann in [Page – 24, Paragraph – 3] discloses, “Tracklet having a candidate with the lowest matching criterion are merged first. Then a merged candidate is removed from any other set of candidates it may belong to. In other words, the step of merging tracklets with its best candidate is done sequentially in increasing order of the value of the comparison score between each tracklet and its best candidate”).
It would have been obvious to one with one having an ordinary skill in art before the effective filling date of the claimed invention to integrate the technique of Johann into the system of Peng in view of Barburescu and Jeon would reduce incorrect association and improving the accuracy of cross camera target identification.
Summary of Citations (Johann)
[Page – 23, Paragraph – 1]; “In a step 1320, for each tracklet a set of candidates for merging is selected”.
[Page – 24, Paragraph – 3]; “Tracklet having a candidate with the lowest matching criterion are merged first. Then a merged candidate is removed from any other set of candidates it may belong to. In other words, the step of merging tracklets with its best candidate is done sequentially in increasing order of the value of the comparison score between each tracklet and its best candidate”.
Summary of Citations (Peng)
[Section – 3.3.4, Paragraph – 7]; “Tracklet clustering aims to associate all the tracklets except the Disappeared ones ... At first, distance matrix MT −cluster is constructed, then a greedy algorithm is adopted to associate the tracklets with the distance threshold
θ
c
l
u
s
t
e
r
”.
Summary of Citations (Jeon)
Paragraph [0030]; “a step of eliminating erroneously detected tracklet may be performed. In other words, to reduce the pool of objects to determine if the tracking of the object succeeds or fails, it eliminates the badly detected, unreliable, traffic. A non-stable tracklet refers to a tracklet having a length less than a predetermined threshold”.
Paragraph [0032]; “For this purpose, it is possible to delay a predetermined number of frames, and through this delay, it is possible to remove a tracklet which appears only in frames less than a predetermined number”.
Paragraph [0049]; “it is possible to distinguish between the tracklets that appear in all nine frames and the tracklets that appear only in less than nine frames, leaving only the tracklets that appear in all nine frames, and the tracklets that appear only in less than nine frames”.
Paragraph [0051]; “the remaining Tracklet can be further classified into two categories. That is, it can be classified as an alive tracklet, which is a stable tracklet but a track that has failed to track before the current frame, which is a dead tracklet and a stable tracklet that has been traced in the current frame”.
Paragraph [0054]; “is possible to associate a dead tracklet with ID1 with a live tracklet with ID3, and reassign the ID of the associated object. That is, it reallocates ID3 to ID1”.
Regarding claim 3, the same grounds of rejection based on Jeon from the last Office Action (Non-Final, 07/30/2025) is incorporated herein.
Regarding claim 11, apparatus claim 11 corresponds to method claim 2. Therefore, the rejection analysis and motivation to combine of claim 2 is applicable to claim 11 as set forth in the last Office Action (Non-Final, 07/30/2025).
Regarding claim 12, apparatus claim 12 corresponds to method claim 3. Therefore, the rejection analysis and motivation to combine of claim 3 is applicable to claim 12 as set forth in the last Office Action (Non-Final, 07/30/2025).
Regarding claim 13, Peng in the combination discloses the device according to claim 9, wherein filtering the tracklets in the set comprises: a) sequentially determining whether to add each tracklet in the tracklet sequence into one or more candidate sets (Peng in [Section – 3.3.4, Paragraph – 7] discloses, “Tracklet clustering aims to associate all the tracklets except the Disappeared ones ... At first, distance matrix MT −cluster is constructed, then a greedy algorithm is adopted to associate the tracklets with the distance threshold
θ
c
l
u
s
t
e
r
”).
The combination of Peng and Barburescu doesn’t disclose the following limitations as further recited in the claim.
Jeon discloses b) removing an earliest tracklet in the tracklet sequence to truncate the tracklet sequence (Jeon in [0032] discloses, “it is possible to remove a tracklet which appears only in frames less than a predetermined number”); c) iteratively repeating step b) until a value obtained by subtracting 1 from a number of tracklets in a current truncated tracklet sequence is less than or equal to a maximum number of tracklets in any of the one or more candidate sets (Jeon in [0030] further discloses about truncating the number of tracklets, “a step of eliminating erroneously detected tracklet may be performed. In other words, to reduce the pool of objects to determine if the tracking of the object succeeds or fails, it eliminates the badly detected, unreliable, traffic. A non-stable tracklet refers to a tracklet having a length less than a predetermined threshold”); and d) taking the candidate set including the maximum number of tracklets (Jeon in [0054] discloses final step of selecting candidate sets from filtered set, ““is possible to associate a dead tracklet with ID1 with a live tracklet with ID3, and reassign the ID of the associated object. That is, it reallocates ID3 to ID1”).
It would have been obvious to one with one having an ordinary skill in art before the effective filling date of the claimed invention to integrate the technique of Jeon into the system of Peng in view of Barburescu because it would allow the system to perform the function more efficiently avoiding unnecessary comparison.
Summary of Citations (Peng)
[Section – 3.3.4, Paragraph – 7]; “Tracklet clustering aims to associate all the tracklets except the Disappeared ones ... At first, distance matrix MT −cluster is constructed, then a greedy algorithm is adopted to associate the tracklets with the distance threshold
θ
c
l
u
s
t
e
r
”.
Summary of Citations (Jeon)
Paragraph [0030]; “a step of eliminating erroneously detected tracklet may be performed. In other words, to reduce the pool of objects to determine if the tracking of the object succeeds or fails, it eliminates the badly detected, unreliable, traffic. A non-stable tracklet refers to a tracklet having a length less than a predetermined threshold”.
Paragraph [0032]; “For this purpose, it is possible to delay a predetermined number of frames, and through this delay, it is possible to remove a tracklet which appears only in frames less than a predetermined number”.
Paragraph [0054]; “is possible to associate a dead tracklet with ID1 with a live tracklet with ID3, and reassign the ID of the associated object. That is, it reallocates ID3 to ID1”.
Regarding claim 15, apparatus claim 15 corresponds to method claim 13. Therefore, the rejection analysis and motivation to combine of claim 13 is applicable to claim 15 as set forth in the last Office Action (Non-Final, 07/30/2025).
Claim 6 is rejected under 35 U.S.C 103 as being unpatentable over Peng in view of Barburescu and further in view of Fisher US Patent Publication No. US-11948313-B2 (hereinafter Fisher).
Regarding claim 6, the same ground of rejection based on Fisher from the last Office Action (Non-Final, 07/30/2025) applies here.
Claims 7 and 14 are rejected under 35 U.S.C 103 as being unpatentable over Peng in view of Barburescu and further in view of Zhang “ByteTrack: Multi-Object Tracking by Associating Every Detection Box” (hereinafter Zhang) and Tran “Robust Traffic-Aware CityScale Multi-Camera Vehicle Tracking Of Vehicles” (hereinafter Tran).
Regarding claim 7, the same ground of rejection based on Zhang and Tran from the last Office Action (Non-Final, 07/30/2025) applies here.
Regarding claim 14, Barburescu in the combination discloses the device according to claim 9, wherein the tracklet extracted for each target appearing in the image sequence is a set of target boxes that identify the target in a plurality of frames of the image sequence respectively (Barburescu in [0084] discloses, “The track path T.sub.ID={(x.sub.1, y.sub.1), (x.sub.2, y.sub.2), . . . } will represent a vector of spatial coordinates of the centers of the bounding boxes corresponding to the person ID”), and the feature extracted for the tracklet is a set of features that are extracted for the target boxes respectively (Barburescu in [0090] discloses, “The VKD neural network learns a numeric appearance descriptor of a person ...The VKD architecture consists of a Resnet feature extractor e.g. Resnet50 or Resnet101 and a classification head”).
Zhang further discloses the feature of each target box used for calculating the similarity is extracted only from a subset of target boxes that satisfy a plurality of quality conditions (Zhang in [Page – 2; Paragraph – 1 (right side)] discloses, “The similarity can be computed by the IoU or Re-ID feature dis tance of the predicted box and the detection box”. Furthermore, Zhang in [Page – 4; Paragraph – 3 (left side)] discloses, “We separate all the detection boxes into two parts Dhigh and Dlow according to the detection score threshold τ”), the quality conditions including (Zhang in [Page – 4; Paragraph – 4 (right side)] discloses, “We find it important to use IoU alone as the Similarity#2 in the second association because the low score detection boxes usually contains severe occlusion or motion blur and appearance features are not reliable”).
Tran further discloses at least a size of the target box (Tran in [Section – 3.5.1, Paragraph - 1] discloses, “We calculate the features by using the bounding boxes that size is bigger than ψb”).
Summary of Citations (Tran)
[Section – 3.5.1, Paragraph - 1]; “We calculate the features by using the bounding boxes that size is bigger than ψb”
Summary of Citations (Barburescu)
Paragraph [0084]; “the Person Tracker 402c uses the person detector 402b to establish bounding boxes around every person detected in every image of captured video footage. Related to these bounding boxes, unique IDs will be assigned to each detected person. The track path T.sub.ID={(x.sub.1, y.sub.1), (x.sub.2, y.sub.2), . . . } will represent a vector of spatial coordinates of the centers of the bounding boxes corresponding to the person ID”.
Paragraph [0090]; “The VKD neural network learns a numeric appearance descriptor of a person ...The VKD architecture consists of a Resnet feature extractor e.g. Resnet50 or Resnet101 and a classification head”.
Summary of Citations (Zhang)
[Page – 2; Paragraph – 1 (right side)]; “The similarity can be computed by the IoU or Re-ID feature dis tance of the predicted box and the detection box”.
[Page – 4; Paragraph – 3 (left side)]; “We separate all the detection boxes into two parts Dhigh and Dlow according to the detection score threshold τ”.
[Page – 4; Paragraph – 4 (right side)]; “We find it important to use IoU alone as the Similarity#2 in the second association because the low score detection boxes usually contains severe occlusion or motion blur and appearance features are not reliable”.
Claims 8 and 16 are rejected under 35 U.S.C 103 as being unpatentable over Peng “in view of Barburescu and further in view of Zhimeng “Multi-Target, Multi-Camera Tracking by Hierarchical Clustering” (hereinafter Zhimeng).
Regarding claim 8, the same ground of rejection based on Zhimeng from the last Office Action (Non-Final, 07/30/2025) applies in here.
Regarding claim 16, Zhimeng in the combination discloses the method according to claim 8.
Peng, Barburescu and Zhimeng doesn’t disclose the following limitations as further recited in the claim.
Zhang further discloses the feature of each target box used for calculating the similarity is extracted only from a subset of target boxes that satisfy a plurality of quality conditions (Zhang in [Page – 2; Paragraph – 1 (right side)] discloses, “The similarity can be computed by the IoU or Re-ID feature dis tance of the predicted box and the detection box”. Furthermore, Zhang in [Page – 4; Paragraph – 3 (left side)] discloses, “We separate all the detection boxes into two parts Dhigh and Dlow according to the detection score threshold τ”), the quality conditions including (Zhang in [Page – 4; Paragraph – 4 (right side)] discloses, “We find it important to use IoU alone as the Similarity#2 in the second association because the low score detection boxes usually contains severe occlusion or motion blur and appearance features are not reliable”).
It would have been obvious to one with one having an ordinary skill in art before the effective filling date of the claimed invention to integrate the technique of Zhang into the system of Peng in view of Barburescu and Zhimeng because ensuring the similarity is computed using only reliable and high quality visual information will significantly improve the accuracy of multi camera target matching.
Tran further discloses at least a size of the target box (Tran in [Section – 3.5.1, Paragraph - 1] discloses, “We calculate the features by using the bounding boxes that size is bigger than ψb”).
It would have been obvious to one with one having an ordinary skill in art before the effective filling date of the claimed invention to integrate the technique of Tran into the system of Peng in view of Barburescu and Zhimeng because it ensures that features extracted from sufficiently large and informative detection resulting in improving feature reliability.
Summary of Citations (Tran)
[Section – 3.5.1, Paragraph - 1]; “We calculate the features by using the bounding boxes that size is bigger than ψb”
Summary of Citations (Barburescu)
Paragraph [0084]; “the Person Tracker 402c uses the person detector 402b to establish bounding boxes around every person detected in every image of captured video footage. Related to these bounding boxes, unique IDs will be assigned to each detected person. The track path T.sub.ID={(x.sub.1, y.sub.1), (x.sub.2, y.sub.2), . . . } will represent a vector of spatial coordinates of the centers of the bounding boxes corresponding to the person ID”.
Paragraph [0090]; “The VKD neural network learns a numeric appearance descriptor of a person ...The VKD architecture consists of a Resnet feature extractor e.g. Resnet50 or Resnet101 and a classification head”.
Summary of Citations (Zhang)
[Page – 2; Paragraph – 1 (right side)]; “The similarity can be computed by the IoU or Re-ID feature dis tance of the predicted box and the detection box”.
[Page – 4; Paragraph – 3 (left side)]; “We separate all the detection boxes into two parts Dhigh and Dlow according to the detection score threshold τ”.
[Page – 4; Paragraph – 4 (right side)]; “We find it important to use IoU alone as the Similarity#2 in the second association because the low score detection boxes usually contains severe occlusion or motion blur and appearance features are not reliable”.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Contact
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ZAID MUHAMMAD SALEH whose telephone number is (703)756-1684. The examiner can normally be reached M-F 8 am - 5 pm ET.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Vu Le can be reached on (571)272-7332. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272- 1000.
/ZAID MUHAMMAD SALEH/
Examiner, Art Unit 2668
2/08/2026
/VU LE/Supervisory Patent Examiner, Art Unit 2668