Method of quality control of ophthalmic lenses

Abstract

The invention relates to a method for quality checking ophthalmic lenses in a manufacturing system comprising a plurality of successive stations, wherein the lenses are checked by means of a first control station, the method comprising: providing a first training database, a first classification unit, a first decision unit associated with the first control station, classifying a first irregularity on the ophthalmic lens and assigning a label from the first training database by means of a classification function; generating a first result value comprising a positive evaluation, a negative evaluation or a rework evaluation; providing at least one second control station downstream of said first station; providing a second training database, a second classification unit, a second decision unit associated with said second control station; generating a second result value comprising a positive evaluation, a negative evaluation or a rework evaluation; comparing said first result value with said second result value and generating feedback information from said comparison and introducing said feedback information into said first training database and said classification function.

Description

Technical Field

[0001] This invention relates to a method for quality control of ophthalmic lenses in a manufacturing system comprising multiple consecutive stations for processing ophthalmic lenses, wherein quality control of the ophthalmic lenses is inspected by means of at least a first control station by generating a base image of the ophthalmic lens, wherein at least one possible abnormality on the ophthalmic lens is identified. The invention also relates to a system for quality control of ophthalmic lenses using this method. Background Technology

[0002] WO 2023 / 041659 A1 describes a method for quality control of ophthalmic lenses. The quality control described therein involves the identification and classification of possible anomalies on ophthalmic lenses, thereby also testing lenses without anomalies, for example, as part of 100% quality control. Therefore, as stated, anomalies "may" exist, but they do not necessarily occur in every ophthalmic lens.

[0003] Possible anomalies in the underlying image recorded by an ophthalmic lens may involve defects, but also intentional structures such as markings, such as structures present in the mold, and contamination, which may be more or less obvious depending on the anomaly.

[0004] Measurement methods used for quality control include a light source for generating a sinusoidal fringe pattern, which is recorded, for example, by a camera, whereby light generated by the light source passes through an ophthalmic lens and is subsequently recorded by the camera. The sinusoidal fringe pattern is then displayed, for example, on a screen and / or evaluated using appropriate equipment. The sinusoidal fringe pattern may exhibit a phase shift, have different wavelengths, and extend, for example, horizontally and vertically. Using other known methods, certain anomalies can be made more visible. The resulting underlying image is then stored in a database. Different types of anomalies are classified, allowing decisions to be made about which category an anomaly should be assigned to.

[0005] When quality control is performed by a control station before or after the first or more stations in a manufacturing system, shortcomings in quality control arise in manufacturing systems with multiple stations for processing and handling large quantities of lenses. Consequently, the processing or handling of ophthalmic lenses after other stations in the manufacturing system can either mitigate or aggravate anomalies. It is possible that some anomalies from the results of the first control station could prevent the ophthalmic lens from being fed to other stations in the manufacturing system at all, even though this is practically possible because, for example, a subsequent coating on the surface of the ophthalmic lens could eliminate the anomaly or at least classify it as a good part.

[0006] On the other hand, if minor anomalies are assigned to a category, and the lens is fed to a further process according to that category, this may result in the lens being classified as a defective part, especially after coating or other surface treatments. This is contrary to the evaluation from the first evaluation unit, where even simple rework, such as rework performed by a station prior to the first control station, can ultimately produce the lens as a good part. Special manual stations can also be set up, or the manual operation station may not be uniformly constructed with the station directly preceding the main processing station for ophthalmic lens cleaning. However, not every lens rated as defective can be sent back for rework from the second control station. This is particularly due to the fact that once a coating has been applied, the lens generally cannot be returned to any previous station in the manufacturing system.

[0007] Therefore, it is desirable to improve the quality control procedures so that the evaluation of an ophthalmic lens as a good component at the first control station is as close as possible to the evaluation of the lens as a good component after it has undergone the entire processing through the manufacturing system. In this regard, the deviation between the evaluation from the first control station and the final evaluation after the lens has undergone further processing through the entire manufacturing system should be as small as possible. Summary of the Invention

[0008] The object of this invention is to improve a method for quality control of ophthalmic lenses having at least one control station, wherein the method operates in such a way that the evaluation results of the ophthalmic lens from the quality control at the first control station can achieve the most accurate possible prediction of the quality outcome of the ophthalmic lens after it has passed through the entire manufacturing system. Therefore, the evaluation, classification, and / or categorization of ophthalmic lens anomalies from the first control station should be as accurate as possible.

[0009] The object of the present invention is achieved by the method according to claim 1 and the system having the corresponding features according to claim 12. Advantageous improvements of the invention are given in the dependent claims.

[0010] To achieve the objectives of the present invention, the following additional steps are proposed for the method: providing a first training database storing labels containing information including various first anomalies; providing a first classification unit connected to the first training database, wherein the classification unit is trained using the training database; providing a first decision unit connected to a first control station and connected to the classification unit; classifying the first anomalies of the base image on the ophthalmic lens and assigning labels to the ophthalmic lens by means of a classification function; generating a first result value from the classification of the first anomalies, including information on a positive assessment, a negative assessment, or a rework assessment; providing at least one second control station after a first, second, or additional station of the manufacturing system; providing a second training database storing labels containing information including various second anomalies; providing a second classification unit associated with the second training database; providing a second decision unit connected to the second control station and the second classification unit; generating a second result value from the classification of the second anomalies, including information on a positive assessment, a negative assessment, or a rework assessment; comparing the first result value with the second result value; generating feedback information from the comparison and inputting the feedback information into the first training database and the classification function.

[0011] The base images are processed base images, and from sinusoidal images, typically 24 to 100 individual images, several base images are generated, for example, 4 base images. These 4 base images are then further processed, using various filters, and in this way, approximately 30 images can be created, from which various anomalies appear, or anomalies can be identified from combinations of the 30 images.

[0012] Regarding the labels, it should be noted that labels can be assigned to each lens, but preferably, labels can be assigned to each anomaly, certain features of the anomaly, or even some or each pixel from the evaluated anomaly. In this way, each pixel can be assigned a label for each type of anomaly, either within or outside the classification, including an assessment of the anomaly intensity, which then refers to the contiguous region of the classification.

[0013] The result can also depend on the interaction of all the abnormalities; while a single abnormality may lead to, for example, a negative assessment, it is also possible that it may lead to a negative assessment only when combined with another abnormality that does not on its own.

[0014] The core idea of ​​this invention is to implement a method for quality control of ophthalmic lenses, featuring feedback between a second control station and a second decision-making unit, so that results from the second decision-making unit at the first control station can be used by the first decision-making unit. This feedback system extends the process in a learning-capable manner, allowing the quality of information in the training database, particularly labels and their classifications, to be continuously improved in further applications of the process, in conjunction with a classification function.

[0015] Indications that decision-making units, classification units, training databases, and / or control stations are interconnected involve data exchange between these facilities.

[0016] Using at least two control stations according to the invention, a testing infrastructure can be established, in which data from the first control station can also be obtained at a later time point, and the classification of anomalies can also be obtained at a later time point, so that measurement data is stored for later inspection, but not all measurement data is subsequently transferred to the training database. In the case of bias in the evaluation of ophthalmic lenses at the second control station due to defects already present during the inspection at the first control station, this bias can be optimized by adjusting the evaluation algorithm, which, for example, uses a classification function to classify anomalies at the first control station.

[0017] In this way, the classification function can be trained in such a manner that it is improved to eliminate any bias between the evaluation of anomalies from the first control station and the second control station. This is further improved, particularly by storing information in the training database. In this method, the training database can receive additional information, or existing information can be overwritten, and thus its quality is improved, allowing the classification unit to better classify anomalies in ophthalmic lenses, aiming to obtain results from the second control station that are as similar as possible to those from the first control station, ensuring a match between the two result values. The newly acquired information can be stored using labels that have been overwritten or newly stored in the training database.

[0018] Therefore, a fundamental feature of this invention is that feedback information is used to train the classification function and / or create and / or improve labels from the training database in such a way that the decision of the first decision unit regarding whether an ophthalmic lens is a good part, a defective part, or a reworkable part is improved. For example, the first decision unit might classify the ophthalmic lens as a good part, but the assessed anomaly is such that, after passing through another station in the manufacturing system, the classification by the second decision unit after measurement at the second control station gives the ophthalmic lens a negative assessment. The negative assessment from the second decision unit is compared with the assessment from the first decision unit that the lens is a good part, so that the lens can be marked as a defective part or at least a reworkable part for future subsequent measurements. It should be noted that the reclassification by the first decision unit does not necessarily occur immediately; for example, it can be said that only after a corresponding bias regarding a certain anomaly has accumulated can the label be rewritten or recreated so that the first decision unit can differently determine whether the corresponding anomaly is repeated.

[0019] Therefore, the second result value from the second decision unit classifies the lens as a good part, a defective part, or a reworkable part. A good part either originates from a good part from the first decision unit, resulting in a true positive assessment, or it may originate from a defective part from the first decision unit, resulting in a false negative assessment. A defective part either originates from a good part from the first decision unit that was misclassified, resulting in a false positive assessment, or the defective part may originate from a defective part from the first decision unit that was fed into the manufacturing system for further processing, since it cannot be ruled out that it was through other stations, particularly through coatings or other surface treatments. Therefore, the initial negative assessment of the anomaly by the first decision unit will ultimately lead to a good result from the second decision unit, because further processing and handling steps of the ophthalmic lens in other stations after the first control station can reduce or even completely eliminate the anomaly. Anomaly assessment refers to the result of the first station. Finally, it is also possible that the lens is classified as good at the first control station, but a new defect is added for the first time in the further process, causing the lens to be classified as defective at the second control station.

[0020] The classification function of the first decision unit is adjusted according to specific advantages to ensure minimal deviation between the first and second result values. In the first decision unit, anomalies detected by the first control station are compared with anomalies stored in the first training database, thereby assigning the anomalies to a specific category; this is performed by the classification unit. If the detected anomaly corresponds to a specific anomaly stored in the training database, the ophthalmic lens is classified accordingly and a label is provided, including classification data. The label may also include metadata, particularly to ensure that lenses with the corresponding anomalies can be identified even during subsequent passage through the manufacturing system stations, especially by a subsequently set second classification unit or second decision unit.

[0021] A second anomaly that persists after passing through the manufacturing system's stations is marked by a second classification unit, enabling the second decision unit to consistently output a second result value. If, for example, a comparison with labels from a second training database shows a very high match, the classification of the second anomaly is examined to ultimately determine whether it is a good part, a defective part, or a reworkable part, because if certain anomalies exist, lenses that have already passed through the coating station may also be reworked.

[0022] The classification function of the first decision unit is modified in such a way that the negative assessments are proportionally reduced, and more ophthalmic lenses are subsequently fed to a second or other station in the manufacturing system until the proportion of defective parts from the second result value increases, and this increase is adjusted by feedback information to counteract the change in the classification function of the first decision unit. In other words, the change in the classification function can mean an overshoot of the second result value, causing the proportion of defective parts to increase again, thus allowing the classification function to be optimized so that the number of defective parts alternates around the optimal value. Therefore, the first decision unit can screen out too many lenses by making false negative assessments, and without feedback, it cannot approximate the actual scrap rate. Similarly, too few ophthalmic lenses can be sorted out, resulting in false positive assessments, which also causes the assessed scrap rate to deviate further from the actual scrap rate. However, using the method according to the invention, the number of lenses sorted out too many, i.e., false negative assessments, can be reduced, as can the number of lenses sorted out too few (corresponding to false positive assessments). In this way, the assessed scrap rate can approximate the actual scrap rate, thereby reducing the number of lenses generated by the second decision unit as defective parts or parts requiring rework. This improved the quality of the examination, and the second result value led to a higher proportion of positive evaluations for ophthalmic lenses.

[0023] As already mentioned, it is possible that ophthalmic lenses with a negative assessment from the first decision unit can then be assessed as positive and manually transferred to the next station. These lenses include, for example, lenses classified as waste or non-reworkable by the decision unit, but still transferred by the operator to the next station in the manufacturing system. In this case, it is possible that, particularly due to the effects on the lenses at another station, such as through coatings or other surface treatments, anomalies are assigned a new category, according to which the second decision unit subsequently performs a positive assessment. Ophthalmic lenses manually fed to the next station are assessed by the second decision unit, where the second result values ​​of these lenses are also fed back to the classification function of the first decision unit via feedback information. This makes the assessed rejection rate even closer to the actual rejection rate.

[0024] The first decision-making unit can be operated with human assistance or completely autonomously; preferably, it can be operated purely through computer digital means. However, the second decision-making unit can also be operated fully automatically using computer digital means; preferably, in addition to computer digital calculations, the second decision-making unit is operated manually under further human supervision. It is also conceivable that the first and / or second decision-making units operate with the aid of AI modules, i.e., including artificial intelligence, particularly for training classification functions based on feedback information. In particular, artificial intelligence can help optimize the quality of labels from a large amount of measurement data and / or a large number of different anomalies, allowing the first classification unit to create a foundation for the decision-making unit to better compare anomalies with the stored labels in the training database, thereby optimizing the first decision and further improving the quality of the first result value of the first decision-making unit.

[0025] Ophthalmic lenses are directly or at least indirectly labeled, for example, in conjunction with an identification number generated by the manufacturing system, wherein the label contains at least data for identifying a specific ophthalmic lens and data relating to at least one of the first and second anomalies or outcome values. In particular, so-called metadata may be included in the label to identify the lens at a later point in time and to retrieve the outcome value from the decision-making unit.

[0026] The object of the invention is also achieved by a system for quality control of ophthalmic lenses, designed for use in a manufacturing system comprising multiple consecutive stations for processing lenses. The system provides: a control station and a first decision unit are provided, and independently of this, at least a second control station and a second decision unit are provided, wherein the at least second decision unit is connected to the first decision unit via a communication channel, particularly a cloud system, through which feedback information can be transmitted from the second decision unit to the first decision unit to influence the classification function of the first decision unit. Further features and advantages described in conjunction with the method also apply to the system for quality control of ophthalmic lenses according to the invention.

[0027] In particular, at least the first decision unit and / or the second decision unit may include an AI module to utilize artificial intelligence; and it is also conceivable that the first decision unit may be computer-operable and the second decision unit may be computer-operable and / or manually operable. Attached Figure Description

[0028] Other measures to improve the present invention will now be described in more detail with reference to the accompanying drawings and the description of preferred embodiments of the invention. It shows:

[0029] Figure 1 This is a schematic diagram of the system architecture for implementing the method for quality control of ophthalmic lenses according to the present invention;

[0030] Figure 2 The diagram shows the central decision-making unit, which is pierced by an unusual ophthalmic lens, which is schematically shown.

[0031] Figure 3 It is a graph of the scrap rate over time when there is no feedback between the second and first decision-making units; and

[0032] Figure 4 This is a schematic diagram showing the rejection rate over time as a function of the feedback between the second decision unit and the first decision unit according to the present invention. Detailed Implementation

[0033] Figure 1 Several stations 110, 120, 130, and 140 are schematically shown. These stations are shown only as examples and are intended to illustrate components of a manufacturing system 100 for processing and handling ophthalmic lenses 1. A series of ophthalmic lenses 1 are shown being fed to the first station 1110, and in stations 1110 and 120, the lenses 1 are, for example, milled, turned, and polished. Laser engraving or other marking processes may also be performed, and these processes are subsequently subject to quality control.

[0034] The second station 120 is followed by the first control station 10, which is designed, for example, according to the disclosure in WO 2023 / 041659 A1.

[0035] Additionally, stations 130 and 140 may perform processes such as coating, hardening, and other finishing of the surface of the ophthalmic lens 1, such that a final inspection is performed only after these processing steps by means of a second control station 15.

[0036] The first control station 10 generates a base image 11, which is fed to the first classification unit 13 and the first decision unit 14. The first classification unit 13, in conjunction with the first decision unit 14, determines whether the currently tested ophthalmic lens 1 is a positive case P or a negative case N. In a positive case P, lens 1 is transferred to the third station 130, shown as an example. In a negative case, ophthalmic lens 1 is either completely rejected, shown as a lens with crossover, and further negative cases may mean the lens is returned to the previous stations 110 and / or 120, or it is manually reprocessed if ophthalmic lens 1 is reprocessable. In the first decision unit 14, a first result value E1 is generated, based on which a decision is made regarding whether the lens represents a positive case P or a negative case N. For this purpose, the first decision unit 14 may have an AI module, shown as a separate unit in this example, but it may also be part of the decision unit 14.

[0037] The second control station 15 is connected to the second classification unit 17 and the second decision unit 18, which examines the further processed ophthalmic lens 1 and generates a second result value E2, which is used to determine whether the part is good or bad. The second training database and the second training unit are not shown here.

[0038] The second result value E2 is first transmitted to the first training database 12 and the first training unit 21 via feedback information R, specifically using the cloud system 19, so that the parameter PA can be trained here and provided to the classification function KF.

[0039] The classification function KF is coupled to the first decision unit 14, allowing the first decision unit 14 to ultimately determine whether the first result value E1 and the second result value E2 match by evaluating the parameters PA from the training unit 21. If they match, as shown in Comparison V, the decision from the first decision unit 14 is correct; if they do not match, the decision from the first decision unit 14 is incorrect. Feedback information R thus continues from the second result value E2 via the cloud system 19, the first training database 12, and the AI-based first training unit 21 from the AI ​​module back to the classification function KF.

[0040] Feedback includes a comparison of the result values ​​E1 and E2, identifying potential for improvement. This allows new labels E to be generated from classification measurement data from control station 10 and training database 12, which are then incorporated into training database 12. New parameters PA are then generated via training module 21 and forwarded to classification unit 13. The new parameters PA are added to the training database.

[0041] To further facilitate this process, ophthalmic lens 1, which is still classified as negative (N) by the first decision unit 14, is fed to another process, namely, to the third station 130. Manual feeding (MZ) can be performed, allowing the operator to continue feeding the negatively assessed ophthalmic lens 1 to the third station 130. If the ophthalmic lens 1 is classified positive in the second control station 15, which is associated with the second decision unit 18, a false negative assessment is generated. Therefore, the results from the second decision unit 18 include good parts and bad parts, i.e., usable lens 1 and lens 1 that is either defective or reworkable. Thus, good parts can be generated by a true positive assessment (TP) and a false negative assessment (FN), while bad parts can be generated by a false positive assessment (FP) or a true negative assessment (TN).

[0042] The illustration also shows a cloud system 19 through which communication between the second decision unit 18 and the first decision unit 14 can be conducted even over greater distances. Of course, direct connection via cable systems, radio systems, etc., is also possible, especially if the first and second control stations 10, 15 or the first and second decision units 14, 18 are close to each other. A local computer can also be used in this regard. However, the cloud system 19 can be advantageous if, for example, a second check using the second control station 15 is performed remotely from a first check using the first control station 10, such as if the manufacturing system 100 is designed as multiple parts and located in different locations.

[0043] Figure 2 The schematic view first shows a first decision unit 14 and a second decision unit 18, wherein the first decision unit 14 is constructed purely by computer, while the second decision unit 18 may be constructed by computer or equipped with an operator.

[0044] Lens 1 is transferred to a first control station 10, each having an exemplary first anomaly A1. Anomaly A1 can represent defects, intentional structures, markings, structures from the mold, contamination, etc., thereby defects can be classified into many different categories, for example, they can be evaluated based on the quotient of the defect's perimeter and area, determined using measurement methods from the first control station 10, which is upstream of the first classification unit 13 and the first decision unit 14. The same applies to the second control station 15, which inspects the completed lens 1 to ultimately feed the results to the second classification unit 18 and the second decision unit 18.

[0045] These numerous different embodiments of the first and second anomalies are stored as labels E in corresponding training databases 12 and 16. The first training database 12 stores the corresponding numerous different implementations of the first and second anomalies in the form of labels E, which are stored in training database 12 or 16 respectively as categories of comparison values. If an ophthalmic lens 1 is to be classified as having the first anomaly A1, anomaly A1 is considered via a first classification unit 13. Following the first training database 12 is a first training unit 21, in which parameters PA are generated. The first classification unit 13 uses parameters PA to classify the lens 1 under test as having the specific anomaly A1. The training procedure must be distinguished from the actual process. During training, the similarity between anomaly A1 and the parameters stored in labels E is evaluated. Then, during the process itself, it is checked whether the anomalies correspond to these parameters, and they are labeled accordingly.

[0046] Finally, these lenses 1 with corresponding condition assessments are forwarded to further processing, represented as positive case P or negative case N, by the manufacturing system 100 after the first decision unit 14. Further processing in the manufacturing system 100 results in the ophthalmic lens 1 being transferred to the second control station 15, upstream of the second classification unit 17 and the second decision unit 18. The second decision unit 18 is also connected via the second classification unit 17 to a second training database 16 storing the tags E. The parameter PA from the second training unit 22 is passed to the second classification unit 17 in the same manner, such that the comparison process is repeated to ultimately classify the lens 1 from the second decision unit 18 accordingly, where, by way of example, the intermediate lens now also represents positive case P, but the lens from the first decision unit 14 is classified as negative case N. Lenses 1 from the second control station 15 and the second decision unit 18 can also be fed back into the manufacturing system 100 for rework.

[0047] The core of this invention is now the comparison between the results of the first decision unit 14 and the second decision unit 18 and the corresponding result values ​​E1 and E2, such that feedback information R is now generated based on this comparison, and this feedback information R is transmitted to the first training database 12 or the first classification unit 13. This allows the quality of the database to be improved using a first anomaly A1 and the corresponding classification, thereby also allowing additional anomalies A1 to be stored in the first training database 12.

[0048] With the aid of artificial intelligence K1, which is not shown in detail here, the first classification unit 13 can be executed as a learning system during another operating cycle of the control system in order to further improve the classification of the abnormality A1 of the ophthalmic lens 1, and in particular improve the quality of the tag E attached to the lens 1, with the aid of the first classification unit 13 as the system is in operation.

[0049] Figure 3 and Figure 4 The so-called normalized scrap rate nA over time t is shown. The curve represents the assessed scrap rate bA, which alternates with the actual scrap rate rA. The curve of the assessed scrap rate bA is shown, according to... Figure 3 The scrap rate bA assessed without feedback information between the second and first decision units does not approximate the actual scrap rate rA, but according to... Figure 4 The scrap rate bA, assessed with feedback information between the second and first decision units, approximates the actual scrap rate rA. This is achieved by... Figure 4 The feedback information from the application, the first decision unit combined with the classification unit to learn and improve the evaluation, makes the result value of the second decision unit more often consistent with the result value of the first decision unit, represented by the evaluated scrap rate bA which is approximately the true scrap rate rA.

[0050] This invention is not limited to the preferred embodiments described above. Rather, many variations of the presented solution, even with fundamentally different designs, are conceivable. All features and / or advantages arising from the claims, description, or drawings, including structural details or spatial arrangements, whether individually or in multiple combinations, may be essential to this invention.

[0051] List of reference numerals in the attached diagram:

[0052] 1. Ophthalmic lenses

[0053] 10 First Control Station

[0054] 11 Basic Images

[0055] 12 First Training Database

[0056] 13 First classification unit

[0057] 14 First Decision-Making Unit

[0058] 15 Second Control Station

[0059] 16 Second Training Database

[0060] 17 Second Classification Unit

[0061] 18 Second Decision-Making Unit

[0062] 19 Cloud System

[0063] 100 Manufacturing System

[0064] 110 First Station

[0065] 120 Second Station

[0066] 130 Third Station

[0067] 140 Fourth Station

[0068] E-label

[0069] E1 First Result Value

[0070] E2 Second Result Value

[0071] PA parameters

[0072] P positive status

[0073] N-negative condition

[0074] FN false negative assessment

[0075] True positive assessment of TP

[0076] FP False Positive Assessment

[0077] True negative assessment of TN

[0078] R feedback information

[0079] V comparison

[0080] A1 First Anomaly

[0081] A2 Second Abnormality

[0082] KF classification function

[0083] MZ manual feeding

[0084] nA Normalized scrap rate

[0085] rA True Scrapping Rate

[0086] bA assesses scrap rate

Claims

1. A method for quality control of ophthalmic lenses (1) in a manufacturing system (100), said manufacturing system comprising a plurality of continuous stations (110, 120, 130, 140) for processing said lenses (1), wherein, The quality control of the ophthalmic lens (1) is checked by generating a base image (11) of the ophthalmic lens (1) using at least a first control station (10), wherein at least one possible abnormality (A1) on the ophthalmic lens (1) is identified. The method includes at least the following steps: - providing a first training database (12) storing labels (E) containing information including various first abnormalities (A1); - providing a first classification unit (13) connected to the first training database (12); - providing a first decision unit (14) connected to the first control station (10) and the classification unit (13); - classifying the first abnormalities (A1) in the base image (11) on the ophthalmic lens (1) and assigning labels (E) to the ophthalmic lens (1) based on training parameters from the first training unit using a classification function (KF); - generating a first result value (E1) from the classification of the first abnormality (A1), including information on a positive assessment, a negative assessment, or a rework assessment. - Provide at least one second control station (15) after the first (110), second (120), or additional station (140) of the manufacturing system (100); - Provide a second training database (16) storing labels (E) containing information including various second anomalies (A2); - Provide a second classification unit (17) connected to the second training database (16); - Provide a second decision unit (18) connected to the second control station (15) and the second classification unit (17); - Generate a second result value (E2) from the classification of the second anomaly (A2), including information on positive assessment, negative assessment, or rework assessment; - Compare the first result value (E1) with the second result value (E2); - Generate feedback information (R) based on the comparison, and - Input the feedback information (R) into the first training database (12) and into the classification function (KF).

2. The method according to claim 1, characterized in that, The feedback information (R) is used to train the classification function (KF) in such a way as to create and / or improve the labels (E) from the training database (12) so that the decision of the first decision unit (14) regarding whether the ophthalmic lens (1) is a good part (G), a bad part (S), or a reworkable part (N) is improved.

3. The method according to claim 1 or 2, characterized in that, The second result value (E2) indicates whether the ophthalmic lens (1) is a good part (G), a defective part (S), or a reworkable part (N), thereby - The acceptable component (G) either originates from the acceptable component (G) of the first decision unit (14), resulting in a true positive (TP) assessment, or the acceptable component (G) originates from the undesirable component (S) of the first decision unit (14), resulting in a false negative (FN) assessment. and / or one of them - The defective component (S) either originates from the good component (G) of the first decision unit (14), resulting in a false positive (FP) assessment, or the defective component (S) originates from the defective component (S) of the first decision unit (14), resulting in a true negative (TN) assessment.

4. The method according to any one of the preceding claims, characterized in that, The classification function (KF) of the first decision unit (14) is adjusted via the feedback information (R) in such a way that the deviation between the first result value (E1) and the second result value (E2) is minimized.

5. The method according to any one of the preceding claims, characterized in that, The classification function (KF) of the first decision unit (14) is adjusted such that the negative cases (N) of the assessment are reduced proportionally, and then more lenses (1) are fed to the second station (120) or another station of the manufacturing system (100) until the proportion of defective parts (S) from the second result value (E2) increases and this increase is readjusted via the feedback information (R) to the classification function (KF) of the first decision unit (14) to offset the change.

6. The method according to any one of the preceding claims, characterized in that, An ophthalmic lens (1) with a negative case (N) from the assessment of the first decision unit (14) is then assessed as positive and is manually transferred to the other station (130).

7. The method according to claim 6, characterized in that, The ophthalmic lens (1) that is manually transferred to the other station (130) by means of the second decision unit (18) is evaluated, wherein the obtained second result value (E2) of these lenses (1) is fed back to the classification function (KF) of the first decision unit (14) via the feedback information (R).

8. The method according to any one of the preceding claims, characterized in that, A local data network or cloud system (19) is established for data exchange, wherein the first control station (10), the second control station (15), the first decision unit (14) and / or the second decision unit (18) communicate with each other or exchange data, particularly in a decentralized manner, by means of the cloud system (19).

9. The method according to any one of the preceding claims, characterized in that, The first decision unit (14) is operated digitally by a computer, and the second decision unit (18) is operated digitally by a computer and / or manually by a person.

10. The method according to any one of the preceding claims, characterized in that, The first decision unit (14) and / or the second decision unit (18) operate by means of an AI module including artificial intelligence, specifically for training the classification function (KF) based on the feedback information (R).

11. The method according to any one of the preceding claims, characterized in that, The ophthalmic lens (1) is provided with a label (E) directly or at least indirectly, the label (E) containing at least data for identifying the specific ophthalmic lens (1) and data relating to at least one of the first and second anomalies (A1, A2).

12. A system for quality control of ophthalmic lenses (1), designed for use in a manufacturing system (100) comprising a plurality of continuous stations (110, 120, 130, 140) for processing the lenses (1), characterized in that, A first control station (10) and a first decision unit (14) are set up, and independently of them, at least one second control station (15) and a second decision unit (18) are set up, wherein at least the second decision unit (18) is connected via a communication channel, particularly a cloud system (19), via which feedback information (R) can be transmitted from the second decision unit (18) to the first decision unit (14) to influence the classification function (KF) of the first decision unit (14).

13. The system according to claim 12, characterized in that, At least the first decision unit (14) and / or the second decision unit (18) have an AI module including artificial intelligence and / or the first decision unit (14) is computer-digitally operable, and the second decision unit (18) is computer-digitally operable and / or manually operable.

Images