The Supreme People's Court, in its second-instance judgment in an administrative appellate case concerning the invalidation of a utility model patent, held that where a claim is drafted in a form of a product claim but substantially belongs to a claim of computer program module framework, it does not fall under the patentable subject matters of utility model patents. For a product claim that includes both hardware improvements and computer program, where the improvement to the prior art lies in the hardware part and the involved computer program is known, it can be deemed as falling under the patentable subject matters of utility model patents; where the claim involves improvements both to the hardware part and to computer program per se, it generally does not fall under the subject matters of utility model patents. Where the claim solely involves improvements to the computer program per se, it generally does not fall under the patentable subject matters of utility model patents.
A brief summary of the case is as follows:
Title of the utility model patent involved in this case is "Intelligent weighing device based on computer vision technology", and the patentee is Company A. Company B, on the grounds that the patent did not fall under the patentable subject matters of utility model patents and lacked inventiveness, filed a requested for invalidation before the CNIPA. The CNIPA issued a decision to uphold the validity of the patent. Company B was dissatisfied with the CNIPA’s decision and instituted an administrative lawsuit with the Beijing Intellectual Property Court (the court of first instance). The first-instance court dismissed the lawsuit of Company B. Unsatisfied with the judgment of the first instance, Company B filed an appeal.
The court of first instance held that:
The patent relates to an intelligent weighing device based on computer-vision technology, which builds the intelligent weighing device with a weighing platform, a vision sensor, a recognition unit, a recognition-feedback unit, a checkout system and a modeling platform. This architecture gives the intelligent weighing device a hardware foundation to recognize products for settlement by using visual information, weight information, recognition models and user feedback, and to return the user feedback to the modeling platform. The program involved in the hardware s already-known computer program existing in the prior art. Compared with conventional scales, the improvement lies in the hardware configuration and the mutual connection relationships among the hardware components, rather than in computer programs. Secondly, the user-feedback data are acquired by the recognition-feedback unit, which may be an intelligent display or a human-machine-interaction device, etc. The results of human judgment and feedback are received through hardware such as intelligent display or a human-machine-interaction device, and then transmitted to a price-label printer or a POS checkout system and the modeling platform. Improvement of claim 1 of the patent lies in the connection relationships among entities rather than addition of computer programs. The technical scheme is to define hardware components and their connection relationships. The essence of the technical scheme is not to implement an information-flow process via computer software. The claimed subject-matter complies with Article 2(3) of the Patent Law and belongs to patentable subject matters of a utility model.
In its appeal against the first-instance judgment, Company B argued that: (I) The patent does not belong to patentable subject matters for utility model patents. What the patent seeks to protect is essentially an improvement to a method. The patent has two objectives: first, to use visual information and weight information to recognize goods through a recognition program and obtain recognition results; second, to increase the recognition model trained with user feedback data and to achieve continuous iterative optimization of the recognition model. Both of the objectives of are achieved through improvements to the computer program, rather than by defining the hardware components and the connection relationships among them. Claim 1 only defines the information transmission and processing relationships between the weighing platform, visual sensor, recognition unit, recognition-feedback unit, checkout system, and modeling platform, and is a typical method claim. (II) Claims 1-8 of the patent lack inventiveness.
The second-instance judgment of the Supreme People's Court overturned the first-instance judgment and the sued decision, ordering the CNIPA to make a new decision on the invalidation request of Company B.
Specifically, the Supreme People’s Court held that:
Claims that are formally drafted as product claims but in substance relate to a framework of computer-program modules do not fall within patentable subject-matters of utility-model patents. To determine whether the technical scheme sought to be protected by a claim qualifies as protectable subject matter for a utility model, the scheme should be assessed as a whole, taking into account the technical means it employs, the technical problem it solves, and the technical effect it achieves.
Firstly, claim 1 of the patent seeks to protect an intelligent weighing device based on computer-vision technology, which comprises both hardware and software portions. Specifically, the software portion is mainly present in two aspects: Aspect 1, claim 1 of the patent defines "a recognition-feedback unit configured to combine the goods recognition result with goods information, display candidate goods information, receive user feedback, transmit the goods information confirmed by the user feedback to a price tag printer or a checkout POS system, and feed it back to a modeling platform, wherein the price tag printer is configured to print a goods weighing barcode based on the user feedback, and the checkout POS system is configured to generate goods payment information based on the user feedback; the modeling platform configured to train a recognition model based on collected visual information, weight information, and user feedback data." According to this definition, whether it is the recognition-feedback unit receiving user feedback, transmitting information to the printer or system, and feeding it back to the modeling platform, or the modeling platform training the model based on the data confirmed by the user feedback, what is substantially defined is the data transmission and connection relationship. And such transmission of these image and text data requires a specific computer program, and cannot be achieved solely through the connection between the hardware components such as the recognition-feedback unit and the modeling platform. Therefore, this definition is essentially a definition on the computer program module per se.
Secondly, according to the description of the patent, it can be seen that the technical solution of this patent essentially combines weighing information, visual information, and goods information confirmed by user feedback as data for a modeling platform. It uses machine learning to build and optimize the model to provide an intelligent weighing device that improves weighing and settlement efficiency. In other words, it primarily achieves its objectives by defining the computer program per se.
Finally, the court of second instance held that, for product claims relating to both hardware improvements and computer programs, if the improvement over the prior art lies in the hardware and the computer program involved is already known, the subject matter is patentable as a utility model. Where the claim involves not only hardware improvements but also improvements to the computer program per se, it generally does not fall within the patentable subject matter of a utility-model patent.
To sum up, the patent at issue does not qualify as patentable subject matter for a utility model. The findings of the first-instance judgment and the sued decision were erroneous and are hereby corrected by the court of second instance.
In this case, regarding the definition on the software portion of the patent, there is no evidence to demonstrate that the combination of weighing information, visual information, and goods information confirmed by user feedback as data for the modeling platform, and the use of machine learning to build and optimize the model, is a known computer program. Furthermore, the sued decision and the first-instance judgment, in discussing inventiveness, held that the distinguishing feature between claim 1 and Evidence 1, "in claim 1 the recognition feedback unit feeds back the goods information confirmed by user feedback to the modeling platform; the modeling platform is configured to train the recognition model based on the collected weight information and user feedback data in use", is not disclosed in Evidence 1 or Evidence 2. Feeding goods information confirmed by user feedback, including image and text data, etc., to the modeling platform necessarily requires a computer program for transmission. This directly contradicts the findings of the sued decision and the first-instance judgment that the program involved in the patent’s hardware is a computer program already known in the prior art. Consequently, the computer program involved in the patent is not known, or the technical scheme of claim 1 substantially involves improvement to the software per se.
Through the above judgment, the Supreme People’s Court has clarified how to assess the subject-matter eligibility of utility-model patents that combine hardware and software, providing guidance for innovators. Applicants facing the filing of such cases should identify whether the improvement lies principally in the hardware or in the software and, in light of the level of inventiveness, select the appropriate type of patent application.
(2023) Zui Gao Fa Zhi Min Zhong No. 576
If you have any question about the protection of intellectual property rights, please feel free to send us emails. For patent-related matters, please send to info@afdip.com. For trademark/litigation/legal matters, please send to info@bhtdlaw.com.
