pKineto

  • 2020-06-21
  • Kickoff meeting for Unexplored
  • Participants: All members of the project led by PM Inami

  • The following is about kineto.

  • We have limited space for writing when it comes to class videos, right?

  • We might connect with HakoScope (an insta360 distributor).

  • How should we store video data? Maybe using AWS?

  • The issue of having two different approaches

    • Is this really a separate axis?
  • Are we aiming for a private tutor-like experience?

    • Is one-on-one tutoring ideal? No, we also want to have discussions with multiple people.
    • It should be possible to ask friends for help when we don’t understand.
  • When multiple people are delayed at the same pace, it might be a good idea to connect those who are falling behind.

  • “Social Note-taking”

    • Since everyone is drawing on the same timeline, we might be able to share notes in a Nico Nico-like fashion.
  • It’s okay to naturally fall behind while taking notes, so a modeless approach is good.

  • When it comes to that, “do we really want to manipulate time with kineto?”

  • I wrote various things about it in What do you really want to manipulate with kineto, time?

  • Interesting experience with drinami

    • When participating in a Zoom meeting and then entering a real room in the middle
    • The sensation of lag being resolved is interesting.
  • Discussion about “Research-like Development”

  • Using people for user studies


  • The following are things that should not be exposed about others.

  • Making emojis a universal language

  • Emoji input method 1 - Select from categories

    • It’s good but difficult (3000 choices).
  • Input method 2 - Conversion during input

    • The choices are narrowed down, there is no randomness.
    • It doesn’t exist in English-speaking countries.
  • We want to provide options while showing the best candidates.

  • We want to avoid bias.

  • Representing emojis in the same space as Japanese using word2vec

    • Challenge: Not being able to understand sentences
  • Building models using BERT and EmojiNet

  • Emojis can be interpreted in multiple ways

    • Mapping multiple vectors

image

  • Creating a graph and suggesting the emoji that best matches

  • It would be nice to express emotions through continuous use of emojis.


  • Umpire training system

  • Distinguishing between strikes and balls

  • The problem is the error rate

    • It’s especially a problem for amateurs
    • Machines that can determine whether it’s an error or not are expensive, and umpires can’t even practice
  • There was a discussion about how spectators express dissatisfaction with incorrect calls. Based on what information do spectators determine whether it’s an incorrect call or not?

  • I don’t know much about baseball, but personally, I was surprised that humans are making these judgments.

  • The benefits of humans making judgments in baseball

    • The culture of framing would disappear, so machine judgments are not good.
  • In fencing, humans can’t determine the outcome with their own eyes

    • If humans can’t determine the outcome, the value of cheering decreases
    • It’s an important point in sports.

  • There is also a problem that hearing-impaired individuals have difficulty in sports development.

  • The collar on the neck is suitable for haptics, so we can transmit sound through the neck.

  • Question

    • One thing that caught my attention is whether there are any effects on hearing through the neck or if it makes able-bodied individuals even stronger.
  • In the future, for sports as a whole

  • Direction where able-bodied individuals can also enjoy

    • Inclusive
    • Good for business as well
  • Issue of body orientation

    • If only the face is turned to the right, which is the “front” in terms of coordinates?
  • In engineering, it has historically been only “militarily useful”

  • You don’t have to worry so much about whether what you’re creating is “useful” in society.


Overall message

  • We should delve into the moments that make us go “wow.”- Example: In a video delay demo, there are moments when the delayed video sticks together.
  • Simply packing it logically may be sufficient for a paper, but it tends to become boring.
  • Emotions are also important.
  • They can also serve as a source of motivation for development.