Prosecution Insights
Last updated: November 04, 2026
Application No. 18/718,488

THERMAL FACE AND LANDMARK DETECTION METHOD

Final Rejection §103
Filed
Jun 11, 2024
Priority
Dec 14, 2021 — EU 21306773.9 +1 more
Examiner
BEZUAYEHU, SOLOMON G
Art Unit
2674
Tech Center
2600 — Communications
Assignee
Inria
OA Round
2 (Final)
76%
Grant Probability
Favorable
3-4
OA Rounds
10m
Est. Remaining
99%
With Interview

Examiner Intelligence

Grants 76% — above average
76%
Career Allowance Rate
490 granted / 645 resolved
+14.0% vs TC avg
Strong +29% interview lift
Without
With
+28.9%
Interview Lift
resolved cases with interview
Typical timeline
3y 2m
Avg Prosecution
45 currently pending
Career history
691
Total Applications
across all art units

Statute-Specific Performance

§101
17.2%
-22.8% vs TC avg
§103
52.3%
+12.3% vs TC avg
§102
12.8%
-27.2% vs TC avg
§112
10.2%
-29.8% vs TC avg
Black line = Tech Center average estimate • Based on career data from 645 resolved cases

Office Action

§103
DETAILED ACTION Response to Arguments Applicant's arguments filed with respect to art rejection of claims 1-4 have been fully considered but are moot in view of the new ground(s) of rejection. The rejections are necessitated due to claim amendments by inserting a condition for performing certain steps. Applicant’s arguments, see page 2-4, filed 8/5/2026, with respect to 101 rejection have been fully considered and are persuasive. The rejection of claims has been withdrawn. Claim Rejections - 35 USC § 103 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 and 3 are rejected under 35 U.S.C. 103 as being unpatentable over Wolff et al. (Pub. No. US 2004/0005086) in view of Shivpure et al. (WO 2021/084375) further in view of Zheng (Pub. No. US 2017/0132458). Regarding claim 1, Wolff teaches a thermal face and landmark detection method comprising the steps of: providing a ground truth (known face/feature) reference database [Para. 7 “match/compare the face representation template of an unknown individual with a stored database or gallery of face templates”; Para. 28 “Once both probes (unknown faces) and gallery (know faces) data have been mapped to the feature vector space, classification proceeds by considering the similarity between the probe and each gallery exemplar”. According to Dictionary of Computer Vision and Image Processing by R.B. Fisher © 2005, ground-truth: In performance analysis, the true value, or the most accurate value achievable, of the output of a specific instrument under analysis, for instance a vision system measuring the diameter of circular holes. Ground truth values may be known theoretically, e.g., from formulae, or obtained through an instrument more accurate than the one being evaluated; Therefore, known-face template gallery reads on ground-truth]; capturing a thermal image (face representation template) comprising at least one face [Para. 7 “In the present invention both video imagery sensed in a sub-spectrum of the reflective domain and video imagery sensed in a sub-spectrum of the thermal infrared domain are used to detect a face, create a face representation template, and match/compare the face representation template of an unknown individual with a stored database or gallery of face templates”; Para. 40 “the task of detecting a face at each respective remote location, acquiring the reflective spectrum/thermal infrared imagery of the detected face and reducing to a face representation template is done by the dedicated computer at that remote location.”, and 27]; applying a thermal face restoration filter method (non-uniformity correction (nuc) image pre-processing) to the thermal image (thermal infrared imagery), thereby creating a corresponding improved thermal image (clarify the image) [Para. 21 “Non-uniformity correction (NUC) image pre-processing, illustrated at 221 in FIG. 2, is used to clarify the image as illustrated at 204, and is required for thermal infrared imagery prior to use for face detection and further creation of a face representation template”]; applying a face detector (detector 230) to the improved thermal image creating a corresponding number of cropped improved face thermal images (cropped with corresponding masks) [Para. 27 “When a face is detected by detector 230 (FIG. 2), eyes for the face are then detected at detector 212 and this detector output is used for geometric normalization of the reflective video image as indicated at 214.”; and “These images are then sub-sampled and subsequently cropped with corresponding masks 215 and 225 that remove all but the inner face, thus eliminating the effect of background, hair, etc”]. However, Wolff doesn’t explicitly teach creating landmarked images by applying a landmark detector to improved cropped face image and apply recognition is on the landmarked image. Shivpure teaches creating landmarked images (determine the position of eyes or eyebrows in the facial of the frames) by applying a landmark detector (facial landmark detection algorithm) to improved (image filtering and enhancement techniques) cropped face image (face region segmented in the frames) and [Para. 63, and 65]. It would have been obvious to one of ordinary skill in the art before the effective filing date to modify Wolff’s thermal-infrared face recognition method by incorporating Shivpure’s teaching of applying a landmark detector (facial landmark detection algorithm) to the improved (filtered and enhanced) cropped face (face region segmented in the frames) thermal image in order to create a landmarked (position of eyes or eyebrows) cropped face thermal image and placing the landmark processing state after image enhancement and face region isolation. This modification improves Wolff by localizing facial landmarks within enhanced facial regions, thereby reducing background noise interference and providing explicit spatial landmarks for recognition. Wolff teaches applying face recognition (by comparing) on the cropped face thermal image against the reference database (gallery of images in a database) [Para. 28 “match the probe image with a gallery of images in a database.”; Para. 36 “Creation of the face representation template and subsequent use of this template to compare and match with existing templates for face recognition applications is performed by software operating on computer system 820” and “This software operating on computer 820 also performs the face detection process monitoring when a face is present in the scene of interest, and determines when to acquire imagery of unknown subjects, store imagery, create face representation templates, and perform matching comparisons for specified face recognition applications”]. However, Wolff in view of Shivpure doesn’t explicitly teach applying face recognition on the landmarked cropped face thermal image against the ground truth reference database. Zheng teaches applying face recognition (hamming distance (HD) between a probe face and each gallery faces is calculated) on the landmarked (aligned (by proper scaling rotation, and translation) according to the face landmarks) cropped face (extracted (cropped)) thermal image against the ground truth (pre-existing database (gallery faces of herself/himself)) reference database [Para. 32, 34, 41 and 62]. It would have been obvious to one of ordinary skill in the art before the effective filing date to modify Wolff’s thermal infrared face recognition method, modified by Shivpure, by incorporating Zhang’s teaching to recognize the landmarked cropped face thermal image against the ground truth reference database, so that the cropped, landmark aligned thermal probe face is compared with stored gallery faces. This modification improves Wolff by normalizing probe face geometry around common landmark before database matching, thereby reducing scale, rotation, and translation variation and increasing matching reliability. Regarding claim 3, Wolff in view of Shivpure teaches all claim limitations as stated above. Furthermore, Shivpure teaches wherein the step detecting the face in the thermal image comprises application (uses) of an object detector [Para. 63]. Claim 2 is rejected under 35 U.S.C. 103 as being unpatentable over Wolff et al. (Pub. No. US 2004/0005086) in view of Short et al. (Pub. No. US 2017/0132458) further in view of Shivpure et al. (WO 2021/084375) and further in view of Zheng (Pub. No. US 2017/0132458). Regarding claim 2, Wolff teaches a thermal face and landmark detection method comprising the steps of: providing a ground truth (known face/feature) reference database [Para. 7 “match/compare the face representation template of an unknown individual with a stored database or gallery of face templates”; Para. 28 “Once both probes (unknown faces) and gallery (know faces) data have been mapped to the feature vector space, classification proceeds by considering the similarity between the probe and each gallery exemplar”. According to Dictionary of Computer Vision and Image Processing by R.B. Fisher © 2005, ground-truth: is the true value, or the most accurate value achievable, of the output of a specific instrument under analysis, for instance a vision system measuring the diameter of circular holes. Ground truth values may be known theoretically, e.g., from formulae, or obtained through an instrument more accurate than the one being evaluated; Therefore, known-face template gallery reads on ground-truth]; capturing a thermal image (face representation template) comprising at least one face [Para. 7 “In the present invention both video imagery sensed in a sub-spectrum of the reflective domain and video imagery sensed in a sub-spectrum of the thermal infrared domain are used to detect a face, create a face representation template, and match/compare the face representation template of an unknown individual with a stored database or gallery of face templates”; Para. 40 “the task of detecting a face at each respective remote location, acquiring the reflective spectrum/thermal infrared imagery of the detected face and reducing to a face representation template is done by the dedicated computer at that remote location.” and para. 27]; applying a face detector (detector 230) to the thermal image, thereby detecting a face and creating a cropped face thermal image [Para. 21, and 27]; applying a method to the cropped face thermal image, thereby creating an improved cropped face (preprocessed face image) thermal image [Para. 27 and 29]. However, Wolff doesn’t explicitly teach applying a Gaussian filters method to the cropped face thermal image, thereby creating an improved cropped face thermal image. Short teaches applying a Gaussian filters (difference of Gaussians (GoG) filter) method to the cropped face thermal image (stroke images), thereby creating (generate) an improved cropped face thermal image [Para. 28, 37, 40 and 41]. It would have been obvious to one of ordinary skill in the art before the effective filing date to modify Wolff’s post masking thermal face normalization stage by incorporating Short’s teaching of applying a Gaussian filters to the cropped face thermal image in order to create an improved cropped face thermal image. This modification improves Wolff by accentuating facial details and reducing high and low frequency noise, thereby producing a cleaner thermal face crop for subsequent landmark processing. Wolff in view of Short doesn’t explicitly teach creating landmarked images by applying a landmark detector to improved cropped face image and apply recognition is on the landmarked image. However, Shivpure teaches creating landmarked images (determine the position of eyes or eyebrows in the facial of the frames) by applying a landmark detector (facial landmark detection algorithm) to improved (image filtering and enhancement techniques) cropped face image (face region segmented in the frames) and [Para. 63, and 65]. It would have been obvious to one of ordinary skill in the art before the effective filing date to modify Wolff’s thermal-infrared face recognition method, modified by Short, by incorporating Shivpure’s teaching of applying a landmark detector (facial landmark detection algorithm) to the improved (filtered and enhanced) cropped face (face region segmented in the frames) thermal image in order to create a landmarked (position of eyes or eyebrows) cropped face thermal image and placing the landmark processing state after image enhancement and face region isolation. This modification improves Wolff by localizing facial landmarks within enhanced facial regions, thereby reducing background noise interference and providing explicit spatial landmarks for recognition. Wolff teaches applying face recognition (by comparing) on the cropped face thermal image against the reference database (gallery of images in a database) [Para. 28 “match the probe image with a gallery of images in a database.”; Para. 36 “Creation of the face representation template and subsequent use of this template to compare and match with existing templates for face recognition applications is performed by software operating on computer system 820” and “This software operating on computer 820 also performs the face detection process monitoring when a face is present in the scene of interest, and determines when to acquire imagery of unknown subjects, store imagery, create face representation templates, and perform matching comparisons for specified face recognition applications”]. Wolff in view of Short further in view of Shivpure doesn’t explicitly teach applying face recognition on the landmarked cropped face thermal image against the ground truth reference database. Zheng teaches applying face recognition (hamming distance (HD) between a probe face and each gallery faces is calculated) on the landmarked (aligned (by proper scaling rotation, and translation) according to the face landmarks) cropped face (extracted (cropped)) thermal image against the ground truth (pre-existing database (gallery faces of herself/himself)) reference database [Para. 32, 34, 41 and 62]. It would have been obvious to one of ordinary skill in the art before the effective filing date to modify Wolff’s thermal infrared face recognition method, modified by Short and Shivpure, by incorporating Zhang’s teaching to recognize the landmarked cropped face thermal image against the ground truth reference database, so that the cropped, landmark aligned thermal probe face is compared with stored gallery faces. This modification improves Wolff by normalizing probe face geometry around common landmark before database matching, thereby reducing scale, rotation, and translation variation and increasing matching reliability. Claim 4 is rejected under 35 U.S.C. 103 as being unpatentable over Wolff et al. (Pub. No. US 2004/0005086) in view of Shivpure et al. (WO 2021/084375) further in view of Zheng (Pub. No. US 2017/0132458) further in view of Askat Kuzedeuov (TFW: “Annotated Thermal Faces in the Wild Dataset”). Regarding claim 4, Wolff in view of Shivpure further in views of Zheng doesn’t explicitly teach the step of detecting a face in the thermal image comprises application of an object detector. However, Askat teaches wherein the object detector and the landmark detector are based on Yolo, especially YoloV5. [Section I paragraph 6 “As a baseline, we trained the YOLOv5 [1] object detection and YOLO5Face [2] face detection models on our dataset without changing their architectures and loss functions. The YOLO5Face model is the adapted version of the YOLOv5 model. we employed the YOLOv5 object detection model and its adapted version for face detection YOLO5Face”; section IV “RetinaFace and YOLO5Face face detection architectures use five facial landmarks as an extra supervision signal, while TinaFace is a general object detection model. In this work, we employed the YOLOv5 [1] object detection model and its adapted version for face detection YOLO5Face [2]”]. It would have been obvious to one of ordinary skill in the art before the effective filing date to modify Wolff in view of Shivpure further in views of Zheng to incorporate object detector application YOLOv5 in the system of Askat; because the modification enables the system to improve thermal face and landmark detection by enabling accurate keypoint extraction and recognition in low quality thermal images using enhanced dataset and YOLO-based models. 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 extension fee 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 date of this final action. Any inquiry concerning this communication or earlier communications from the examiner should be directed to SOLOMON G BEZUAYEHU whose telephone number is (571)270-7452. The examiner can normally be reached on Monday-Friday 10 AM-8 PM. If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Oneal Mistry can be reached on 313-446-4912. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300. Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 888-786-0101 (IN USA OR CANADA) or 571-272-4000. /SOLOMON G BEZUAYEHU/ Primary Examiner, Art Unit 2666
Read full office action

Prosecution Timeline

Jun 11, 2024
Application Filed
Jul 14, 2025
Response after Non-Final Action
Apr 01, 2026
Examiner Interview (Telephonic)
May 05, 2026
Non-Final Rejection mailed — §103
Aug 05, 2026
Response Filed
Oct 13, 2026
Final Rejection mailed — §103 (current)

Precedent Cases

Applications granted by this same examiner with similar technology

Patent 12749214
THREE-DIMENSIONAL POSE ESTIMATION USING TWO-DIMENSIONAL IMAGES
3y 2m to grant Granted Sep 29, 2026
Patent 12731374
SAFE CONTROL/MONITORING OF A COMPUTER-CONTROLLED SYSTEM
3y 2m to grant Granted Sep 08, 2026
Patent 12725271
IMAGE QUALITY ENHANCING
4y 2m to grant Granted Sep 01, 2026
Patent 12721537
SYSTEM AND METHOD FOR MEASURING BLOOD FLOW VELOCITY ON A MICROFLUIDIC CHIP
3y 3m to grant Granted Sep 01, 2026
Patent 12718592
INFORMATION COLLECTION SYSTEM, SERVER, AND INFORMATION COLLECTION METHOD
3y 3m to grant Granted Aug 25, 2026
Study what changed to get past this examiner. Based on 5 most recent grants.

Strategy Recommendation AI-generated — please review before filing

Get a prosecution strategy drawn from examiner precedents, rejection analysis, and claim mapping.
Typically takes 5-10 seconds — AI-generated, attorney review required before filing

Prosecution Projections

3-4
Expected OA Rounds
76%
Grant Probability
99%
With Interview (+28.9%)
3y 2m (~10m remaining)
Median Time to Grant
Moderate
PTA Risk
Based on 645 resolved cases by this examiner. Grant probability derived from career allowance rate.

Sign in with your work email

Enter your email to receive a magic link. No password needed.

Personal email addresses (Gmail, Yahoo, etc.) are not accepted.

Free tier: 3 strategy analyses per month