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 .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 03/16/2026 was considered by the examiner.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
Claim(s) 1, 9, 12, 16, 19 and 20 is/are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Francois et al. (US20210319257A1, “Francois 2021”).
With regards to claim 1, Francois (2021) teaches “receiving rock sample imagery of rock” (unknown-sample imagery obtained by imaging device 180 and received by processing system 196/700), Francois (2021) [0022], [0025], [0049], (Figures 1 & 11).
Francois (2021) teaches “generating characterizations of the rock based at least in part on the imagery using one or more machine learning models” (trained deep-learning model assigns rock classes from the imagery and derives rock types, proportions and properties), Francois (2021) [0027], [0029], [0035]–[0040], (Figures 2 & 5).
Francois (2021) teaches “outputting the characterizations of the rock” (method 200 provides properties of the sample as output, and the deep-learning model provides rock type at pixel level as output), Francois (2021) [0027], [0029], (Figure 2). The linked lithology-profile and visualization/output devices 728 corroborate presentation of those results, Francois (2021) [0040], [0046], [0053], (Figures 2 & 11).
With regards to claim 9, Francois (2021) teaches the method of claim 1 for the reasons stated above and imagery “acquired using equipment at a field site” (cuttings imager 180 at the wellsite), Francois (2021) [0018]–[0022], [0025], (Figure 1). It also teaches “the receiving and the generating” using equipment remote from that site: processing system 700 implements the analysis functions, may be remote from the wellsite, and receives/processes the unknown-sample images, Francois (2021) [0046]–[0049], (Figure 11). The expressly disclosed remote-analysis option is selected while acquisition remains at the wellsite.
With regards to claim 12, Francois (2021) teaches the method of claim 1 for the reasons stated above and “implementing transfer learning” by pretraining the image network and fine-tuning it with annotated rock images before applying the resulting model to the unknown sample. The disclosed training and inference workflow, rather than mere possession of an unexplained pretrained model, supplies this addition, Francois (2021) [0017], [0030]–[0037], (Figure 2).
With regards to claim 16, Francois (2021) teaches the method of claim 1 for the reasons stated above and rock sample imagery of cuttings “from a drill bit of a drillstring disposed in a borehole” (bit 144 on drillstring 136 produces cuttings that are returned, collected and imaged by device 180), Francois (2021) [0018]–[0022], (Figure 1).
With regards to claim 19, Francois (2021) discloses the claimed invention according to claim 1, Francois (2021) further teaches that the system implements those operations with “a processor” (712), “memory operatively coupled to the processor” (714/716 and bus 722), and “a network interface” (interface 724 including a network interface card), Francois (2021) [0046]–[0052], (Figure 11).
Francois (2021) also teaches “processor-executable instructions stored in the memory to instruct the system to” perform those operations (code 732, including the trained model, stored in the disclosed memory and executing the sample-analysis workflow). The linkage is express: [0046] connects the hardware to the previously described methods and [0049] identifies unknown rock sample images as received/processed data, Francois (2021) [0046]–[0049], [0053], [0056], (Figure 11).
With regards to claim 20, Francois (2021) discloses the claimed invention according to claim 1, and Francois (2021) further teaches “one or more computer-readable storage media” (storage 730 and/or removable medium 734) comprising executable code 732 that instructs processing system 700 to perform those same methods, Francois (2021) [0046], [0049], [0053]–[0056], (Figure 11). The expressly tangible, non-transitory media satisfy the claimed storage-media category.
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.
Claim(s) 2, 3 and 10 is/are rejected under 35 U.S.C. 103 as being unpatentable over Francois et al. (US20210319257A1, “Francois 2021”) in view of Fan et al. (2020, DOI 10.1155/2020/7462524).
With regard to claim 2, Francois (2021) teaches the method of claim 1 for the reasons stated above, a field site where the sample is cut and imaged (bit/drillstring and imager 180), and mobile computing as an implementation option, Francois (2021) [0018]–[0022], [0047], (Figures 1 & 11).
Francois (2021), however, does not expressly teach that “the receiving, the generating and the outputting are performed by a mobile device at a field site” in that complete allocation.
Fan teaches performing image acquisition, local trained-model inference and result presentation using an Android smartphone in field rock recognition, Fan, (pp. 6–8, §§2.4–2.4.2).
In view of the utility of obtaining rock classifications at the sampling location without laboratory delay, it would have been obvious to a person of ordinary skill in the art at the time the invention was made to implement Francois (2021)’s receiving, generating and outputting on its mobile-computing option using Fan’s local inference and interface approach. Francois’s field cutting and imaging relationship remains; the phone receives the resulting imagery. The predictable result is on-site characterization on a portable device.
With regard to claim 3, Francois (2021) teaches the primary method and field arrangement discussed for claim 2, but does not expressly teach the mobile device being “not in communication with a communication network that extends to a site remote from the field site.” The claim 2 combination is incorporated in full.
Fan teaches local model execution and specifically identifies suitability for field offline operation, Fan, (p. 8, §2.4.2; p. 10, §3.4).
In view of the utility of classifying samples where remote communication is unavailable, it would have been obvious to a person of ordinary skill in the art at the time the invention was made to operate the claim 2 mobile implementation with the model stored locally and without a remote-network connection, as taught by Fan. The predictable result is receipt, inference and output at the field site while disconnected.
With regard to claim 10, Francois (2021) teaches the method and remote receiving/generating of claim 9 for the reasons stated above, and lists smartphones and servers among computing implementations, Francois (2021) [0046]–[0049].
Francois (2021), however, does not expressly tie the full division of work to field equipment comprising “a mobile device” and remote equipment comprising “cloud-based equipment or a remote server.”
Fan teaches field smartphone imagery and discusses client/server exchange with model inference on a high-performance cloud server as a deployment alternative, Fan, (pp. 6–7, §2.4).
In view of the utility of using remote computing resources while retaining portable field image collection, it would have been obvious to a person of ordinary skill in the art at the time the invention was made to use Fan’s smartphone acquisition and cloud/server arrangement for Francois (2021)’s existing remote-analysis embodiment.
The predictable result is field image acquisition by the mobile device and receipt/generation at the remote server. Fan’s stated latency, cost and data-transfer disadvantages favor its local implementation but do not make the disclosed remote alternative inoperative; Francois already supplies an affirmative remote-processing option.
Claims 4, 5, 7, 8 and 17 are rejected under 35 U.S.C. §103 as being unpatentable over Francois et al. (US20210319257A1, “Francois 2021”) in view of Francois et al. (US20190338637A1, “Francois 2019”).
With regard to claim 4, Francois (2021) teaches the method of claim 1 for the reasons stated above, but does not expressly teach “confidence metrics for the characterizations of the rock.”
Francois (2019) teaches a confidence-level indicator for classified rock-image windows and for the image, derived from classifier likelihoods and feature-distance information, Francois (2019) [0066], [0088], (Figures 2A & 9, see 122, 420 & 424).
In view of the utility of distinguishing reliable rock-image predictions from ambiguous ones, it would have been obvious to a person of ordinary skill in the art at the time the invention was made to add Francois (2019)’s classification-confidence evaluation to Francois (2021)’s rock-image workflow. Both evaluate image-derived rock classes; applying an accompanying likelihood/confidence measure provides a predictable reliability output.
With regards to claim 5, Francois (2021) teaches the primary method, and the claim 4 combination supplies confidence metrics as explained above.
Francois (2021), however, does not expressly teach “responsive to one of the confidence metrics being less than a confidence metric criterion, generating a recommendation that aims to increase confidence.”
Francois (2019) teaches that a below-threshold confidence window is displayed and user input is requested; the geologist validates the category, the reference set is updated and the same image may immediately be classified again. The stated purpose is improving classification, Francois (2019) [0067], (Figure 2A, see 124 & return path 126). Notice that the request to review/validate the uncertain classification is a recommendation for an action aimed at greater confidence, not merely a displayed warning.
In view of the utility of directing expert attention to uncertain rock classifications, it would have been obvious to a person of ordinary skill in the art at the time the invention was made to add Francois (2019)’s threshold-triggered review request to the confidence-equipped Francois (2021) workflow. The predictable result is a recommendation to validate an uncertain sample classification and improve the reference information used for subsequent characterization.
With regards to claim 7, Francois (2021) teaches the method of claim 1 and output devices for visualization, Francois (2021) [0053], (Figure 11).
Francois (2021), however, does not expressly teach that “the outputting comprises rendering a graphical user interface to a display device.”
Francois (2019) teaches displaying image windows on a screen and providing a user interface for selecting an image area and assigning a category, with graphical classification outputs, Francois (2019) [0067], [0086]–[0088], (Figure 9).
In view of the utility of allowing the geologist to inspect and correct automated rock-image results, it would have been obvious to a person of ordinary skill in the art at the time the invention was made to render Francois (2021)’s outputs through Francois (2019)’s image-based user interface. The predictable result is visible, interactive presentation of the rock characterizations.
With regard to claim 8, Francois (2021) teaches the primary workflow and sample/classification images, Francois (2021) [0041]–[0044], (Figures 5–8). Notice that the claim 7 combination supplied the GUI.
Francois (2021), however, does not expressly teach the complete GUI arrangement comprising “at least a portion of the rock sample imagery and one or more graphics indicative of one or more of the characteristics of the rock.”
Francois (2019) teaches image 400 as the original sample image with superimposed class colors and a legend identifying the rock categories, Francois (2019) [0087], (Figure 9, see 400, 402 & 410A–410D).
Francois (2019) teaches, see Figure 9, wherein the source’s [0087], [0088] cross-reference to Figure 4 does not change the verified Figure 9 content.
In view of the utility of relating a predicted rock class to its location in the actual sample, it would have been obvious to a person of ordinary skill in the art at the time the invention was made to present Francois (2021)’s imagery and class results in the GUI using Francois (2019)’s overlay/legend arrangement. The predictable result is the claimed combined imagery and characteristic-indicating graphics. This is specific image-to-result correspondence, not an assumption that every GUI includes such an overlay.
With regard to claim 17, Francois (2021) teaches the method of claim 1, expert rock-image annotations for training, and sample depth information, Francois (2021) [0025], [0030]–[0034], [0040].
Francois (2021), however, does not expressly teach the full additional-data relationship for this ground: “receiving data related to a rock sample” and generating the rock characterizations “based at least in part on the data.” The earlier depth-profile anticipation theory is not retained.
Francois (2019) teaches receiving a user-validated rock category or a category assigned to a user-selected image area, adding that labeled image information to the reference set, and running classification again on the same image using the updated set to compute updated results, Francois (2019) [0067], (Figure 2A, see 124 & 126). The received sample-category label affects the generation of the updated characterization; it is not merely appended to an unchanged output.
In view of the utility of incorporating expert correction when image-only rock classification is unreliable, it would have been obvious to a person of ordinary skill in the art at the time the invention was made to add Francois (2019)’s label-feedback and same-image reclassification path to Francois (2021)’s labeled-training workflow. The modification uses the received category data to add the received category label and corresponding sample imagery to the supervised training examples, execute the primary’s training process to update the CNN weights, and then rerun the same sample image using the updated model. Merely adding a record to a database would not update a trained CNN. The predictable result depends on both sample imagery and the received sample-related label. No real-time training deadline or specific feature-fusion architecture is claimed.
Claim 6 is rejected under 35 U.S.C. §103 as being unpatentable over Francois et al. (US20210319257A1, “Francois 2021”) in view of Francois et al. (US20190338637A1, “Francois 2019”), as applied to claim 5 above, and further in view of Androvandi et al. (WO2023230101A1).
With regards to claim 6, Francois (2021) describes field acid testing of isolated cuttings when visual inspection leaves rock identity unclear, Francois (2021) [0002].
Francois (2021), however, does not expressly teach that the responsive recommendation comprises “a test recommendation for performing a test on at least a portion of a rock sample at a field site.”
Francois (2019) teaches the inherited low-confidence review request and same-image reclassification after expert validation, Francois (2019) [0066]–[0067], (Figure 2A, see 122, 124 & 126).
Androvandi teaches an application-mediated request for additional physical or chemical testing of the imaged cuttings at the rig. The workflow transfers images and lithology clusters for expert review and returns requests 190 to rig personnel, Androvandi [0018]–[0019], [0030], [0033], (Figure 5).
Notice that the available tests include hardness, calcimetry on mixed samples or particles of an individual lithology, fluorescence, density and acid reactivity; additional tests may be specified after expert review of the images and clusters, Androvandi [0028].
In view of the utility of obtaining discriminating physical information when visual rock identification remains uncertain, it would have been obvious to a person of ordinary skill in the art at the time the invention was made to extend the low-confidence review branch of Francois (2021), as modified by Francois (2019), with Androvandi’s application-mediated field-test request.
Notice that the below-criterion confidence metric initiate’s expert review; if that review requires physical information to resolve the sample identity, the expert uses the application to generate a recommendation for an appropriate disclosed field test on that sample. The expected result is a recommendation aimed at increasing confidence through additional diagnostic information. The modification preserves the existing image/ML characterization and uses an expressly taught remedy for incomplete image-based identification.
Claims 11 and 15 are rejected under 35 U.S.C. §103 as being unpatentable over Francois et al. (US20210319257A1, “Francois 2021”) in view of Vaidyanathan et al. (US20210358103A1).
With regards to claim 11, Francois (2021) teaches the method of claim 1 and control commands for wellsite preparation/imaging equipment, Francois (2021) [0025], [0049], [0052].
Francois (2021), however, does not expressly teach “controlling field equipment based at least in part on the characterizations of the rock”; commands for preparing an image are not enough to prove that causal relationship.
Vaidyanathan teaches mineral exploration using image-classification module 272 and a field vehicle with navigation module 230, Vaidyanathan [0030]–[0038], (Figure 2).
Notice that the system uses sample-image classifications in determining mineral amounts/proportions at a site and uses negative mineral-content outcomes to change the vehicle’s route and bypass remaining sites in an unpromising subregion, Vaidyanathan [0067]–[0071], [0073], (Figure 9, see 930–970). The image-derived result contributes to the control decision even though XRF information also contributes.
In view of the utility of directing field sampling toward regions with useful mineral content, it would have been obvious to a person of ordinary skill in the art at the time the invention was made to use Francois (2021)’s image-derived rock characterizations as an input to Vaidyanathan’s disclosed exploration-vehicle control logic. The modification applies the primary’s rock-image analysis to an exploration sampling workflow and uses the resulting classes/proportions in the route decision.
The predictable result is field equipment movement controlled at least partly by the characterizations, saving sampling effort in unpromising areas.
With regard to claim 15, Francois (2021) teaches the method of claim 1 and field rock-image acquisition, Francois (2021) [0018]–[0022], (Figure 1).
Francois (2021), however, does not expressly teach that the imagery is “from a field site associated with mineral exploration.” The oil field context is not automatically equated with this added purpose.
Vaidyanathan teaches field exploration of potential mineral-extraction areas, acquisition of outcrop/sample images, their processing by image-classification models, and generation of identified sample properties, Vaidyanathan [0025]–[0035], [0067], [0072]–[0076], (Figures 1, 2 & 10).
In view of the utility of identifying rock characteristics while prospecting for mineral resources, it would have been obvious to a person of ordinary skill in the art at the time the invention was made to apply Francois (2021)’s sample-image analysis to imagery acquired at Vaidyanathan’s mineral-exploration field sites.
The predictable result is rock characterization of imagery from the claimed source context.
Claims 13 and 14 are rejected under 35 U.S.C. §103 as being unpatentable over Francois et al. (US20210319257A1, “Francois 2021”) in view of Alzubaidi et al. (2022, DOI 10.1007/s00603-022-02805-y).
With regard to claim 13, Francois (2021) teaches the claim 12 transfer-learning method for the reasons stated above and training/inference on rock imagery, Francois (2021) [0030]–[0037].
Francois (2021), however, does not expressly teach transfer comprising “a first domain that utilizes pre-existing rock sample imagery and a second domain that utilizes the rock sample imagery.”
Alzubaidi teaches a model trained with sandstone core images and proposes using that trained model to initialize network parameters and retrain on a new dataset with more images or classes, Alzubaidi, (p. 3732, §5.2.4 final paragraph; source-model training at pp. 3724, 3727, Figures 7 & 9). The source’s learned parameters derive from rock imagery; the new rock-image dataset supplies the target domain.
In view of the utility of adapting a rock-image model to additional sample types while reusing learned rock features, it would have been obvious to a person of ordinary skill in the art at the time the invention was made to use Alzubaidi’s rock-trained initialization and target retraining in Francois (2021)’s existing training/inference workflow.
The predictable result is transfer from pre-existing rock imagery to the target rock-image domain, followed by characterization using the adapted model.
With regards to claim 14, Francois (2021) teaches the primary method and transfer-learning foundation, and claim 13 combination supplies the rock-to-rock domains.
Francois (2021), however, does not expressly teach the complete relationship in which knowledge from the first imagery, processed through a first model, is transferred “to a second one of the one or more machine learning models” for those rock domains.
Alzubaidi teaches initializing the target network with parameters of the existing trained model and retraining on the new dataset, Alzubaidi, (p. 3732, §5.2.4 final paragraph). The existing trained model is the first model; the initialized and retrained target is a second model instance. The learned parameter values are the transferred knowledge.
In view of the utility of retaining rock-feature knowledge when building the target model, it would have been obvious to a person of ordinary skill in the art at the time the invention was made to implement the claim 13 transfer as the source-parameter initialization and target retraining that Alzubaidi describes.
The predictable result is a target model that receives knowledge learned by the first model from the first rock imagery.
Claim 18 is rejected under 35 U.S.C. §103 as being unpatentable over Francois et al. (US20210319257A1, “Francois 2021”) in view of Francois et al. (US20190338637A1, “Francois 2019”), as applied to claim 17 above, and further in view of Al-Dhubaib et al. (US20050197841A1).
With regards to claim 18, Francois (2021) teaches the primary method, including mobile computing and optional voice-recognition input devices, Francois (2021) [0047], [0053], (Figure 11). The claim 17 combination supplies the sample-label input that affects regenerated characterization.
Francois (2021), however, does not expressly teach that the relevant data comprise “voice recognition data generated by utterances received by a microphone of a mobile device.” Its generic voice-input list alone is insufficient to establish that content and causal role.
Al-Dhubaib teaches capturing spoken descriptions of rock samples as structured geological data, including mineral terminology and quantitative sample descriptions. Microphone 11 supplies utterances through audio interface 12 to speech-recognition system 13, which produces corresponding data, Al-Dhubaib [0027]–[0035], [0039]–[0044], (Figure 1).
The reason for speech entry is hands-free, eyes-free sample examination in a dirty working environment, Al-Dhubaib [0027]–[0029], [0053].
In view of the utility of entering sample descriptions while keeping the geologist’s hands and attention on the sample, it would have been obvious to a person of ordinary skill in the art at the time the invention was made to implement the claim 17 category-entry path using Al-Dhubaib’s microphone and geological speech-to-data interface on Francois (2021)’s mobile-computing option. The recognized category data would populate the same validated sample-label input used in claim 17 in combination of adding the labeled imagery to training, update CNN weights, and rerun that sample image to regenerate the characterization.
This is a predictable substitution of spoken geological entry for manual geological entry, using the expressly taught voice-input capability and the same feedback destination.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Djura Malevic whose telephone number is 571.272.5975. The examiner can normally be reached M-F (9-5).
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, Uzma Alam can be reached at 571.272.3995. 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.
/DJURA MALEVIC/Examiner, Art Unit 2884
/UZMA ALAM/Supervisory Patent Examiner, Art Unit 2884
Prosecution Insights