This decision relates to a European patent application that concerns the extraction of scenes from a captured video of a sports game. The Board noted that the scenes are defined by characteristics of the audio signal volume based on a heuristic rule reflecting non-technical considerations about the behaviour of spectators at sports games. Thus, the identification of types of scenes of a sports game is not a technical purpose. Here are the practical takeaways from the decision T 2263/18 (Extraction method/FUJITSU) of October 11, 2021, of Technical Board of Appeal 3.5.07:
The application relates to the extraction of scenes from a captured video of a sports game. In known video distribution services, it is sometimes desired to identify scenes of interest from a video, for example, scenes in which a point is scored. It was known to recognise a scene of interest by detecting a sharp rise in sound volume in audio data of the video. The present invention proposes to improve the known video extraction techniques for extracting a scene which is defined by a start point and a continuation time (duration of the scene) in a sports video
Fig. 1 of EP 2921973 A1
Claim 1 (main request)
An extraction program (50) that causes a computer (40) to execute processing comprising:
extracting (54) a portion of video data that corresponds to a continuation time as a scene of interest in a sports game, and determining the type of the scene of interest,
wherein the continuation time is a period of time during which a sound volume level of audio data included in footage including audio data and the video data, captured in a sports game, is detected to have a volume greater than a predetermined level indicative of a scene of interest, and
said extracting comprises comparing the continuation time with each of plural thresholds which have different lengths according to a respective type of scene of interest,
the processing further comprising:
detecting, as a start point, a point in time at which the change amount in the sound volume has exceeded a predetermined change amount threshold; and
determining, as the continuation time, a period from the start point in which the sound volume is continually at a level equal to or higher than a specific level based on the sound volume corresponding to the start point.
Is it patentable?
The first-instance Examining Division found that the subject-matter of claim 1 of this request lacked an inventive step over the disclosure of D2 (US 6,160,950) as the distinguishing features were considered as having no synergistic technical effect. The distinguishing feature over document D2 was identified as follows:
A The type of scene of interest is determined by comparing the continuation time against a variety of thresholds, each associated with a respective type of scene of interest.
B The parameter used to set the start point and the continuation time of a scene of interest is the change amount [in the sound volume].
However, the appellant disagreed and argued the features had a technical effect, as follows:
The appellant argued that feature A solved the problem of identifying scene types by comparing the continuation time with each of a set of thresholds corresponding to different scene lengths. These scene lengths defined respective types of scenes of interest. Thus the claimed type determination was based on numerical comparisons, not on any “cognitive interpretation”. The objective technical problem solved by feature A was how to identify scene types, and the solution was not obvious.
Feature B avoided the incorrect extraction of video portions due to incorrectly calibrated audio levels or if the sound level in an arena recorded in sports footage was continuously high. A relative rise in the sound level could be reliably detected by using the “change amount” of the sound level in detecting the start point. Thus using the “change amount” would contribute to accurate detection of the “continuation time” (statement of grounds of appeal, page 8). The extraction method disclosed in document D2 did not address the problems solved.
Finally, features A and B in combination provided the synergistic effect of improving the accuracy of identified scene types, which was a technical effect.
The Board disagreed with the arguments of the appellant, and agreed with the division that the distinguishing feature does not include a synergistic technical effect:
3.3 The board does not recognise the alleged synergistic technical effect. The board agrees that features A and B interact, but the identified scene types correspond to different types of scenes of a sports game (see for example description, paragraph : goal scenes, disappointment scenes, attack defense switch scenes, foul scenes) and thus relate to the non-technical cognitive content of the video. Hence the combined purpose of features A and B of identifying scene types and their starting points is non-technical.
Moreover, the fact that features A and B interact does not imply that they produce a synergistic effect. Rather, the overall effect is the result of aggregating these features: feature B concerns how the starting point and continuation time are obtained and feature A determines the scene type based on the continuation time. That the scene type is more accurate when the detection of the starting point and the continuation time is more accurate is the expected result of the aggregation of features A and B.
The Board then assessed if the individual features may contribute to the technical purpose that can be used for the assessment of inventive step:
3.4 As the identification of types of scenes of a sports game is not a technical purpose and as the board does not see that feature A involves any “further technical considerations” (see opinion G 0003/08, OJ EPO 2011, 10, Reasons 13.5 and 13.5.1), the board agrees with the examining division that feature A does not contribute to solving a technical problem. Hence it does not enter into the assessment of inventive step (see decision T 154/04, OJ EPO 2008, 46, point 5 (F) of the reasons: “Non-technical features, to the extent that they do not interact with the technical subject matter of the claim for solving a technical problem, i.e. non-technical features ‘as such’, do not provide a technical contribution to the prior art and are thus ignored in assessing novelty and inventive step”).
3.5 The definition of the scenes by means of characteristics of the audio signal volume is based on a heuristic rule reflecting non-technical considerations about the behaviour of spectators at sports games which is usually observed, namely that the sound level increases at time periods of particular interest for the public watching a sports game such as when a goal is scored in soccer. Such time periods then correspond to various types of scenes of interest in the video of the sports game. The fact that the implementation of such a heuristic rule in the system may involve the use of technical means for determining the change amount in the sound volume does not render the underlying heuristic itself technical (see for example decision T 1670/07 of 11 July 2013, reasons 9).
According to the established case law of the boards of appeal, when assessing inventive step in accordance with the problem/solution approach, an aim to be achieved in a non-technical field may legitimately appear in the formulation of the problem as part of the framework of the technical problem to be solved as a constraint that has to be met (see decisions T 641/00, OJ EPO 2003, 352; T 154/04, OJ EPO 2008, 46). Consequently, the heuristic rule may be added as a non-technical constraint to the objective technical problem to be solved.
The board agrees with the examining division that the implementation of the heuristic rule in the program known from document D2 with the aim of determining the type of scenes of interest was straightforward and could not be the basis for acknowledging inventive step. The skilled person was familiar with “change amounts” of sound levels and able to come up with the necessary concepts for implementation.
Therefore, claim 1 was considered not to involve an inventive step, and the appeal is dismissed and the application was refused.
You can read the whole decision here: T 2263/18 (Extraction method/FUJITSU) of October 11, 2021, of Technical Board of Appeal 3.5.07