Building present in the open,
one week at a time.
Thanks for checking out! Moving regular updates to subscribed friends only for now. Will post more as we make more progress. Stay tuned >:)
Detect book cover
Hi friends! I'm finally catching up the slack! ( sentimental stuff here - was slightly lost for a while after knowing the complexity of this project + reading it's time to build. This project can be very useful if done right, else super useless)
Product side - Revisited long term vision. Instead of steer deep into one specific use case, expand interactions to accommodate more possible use cases. (more to come next week)
Hardware side - Stable, even though I am aware now that my pi + pi-camera are crap for real time processing. :(
Software side - Done real time book framing! Turns out it's super basic. Next will be the hard part - make sure it's agnostic to various lighting situation and cover complexity. Need to re-prioritize cover vs inline annotation code next week.
Recruit - Can't do this alone! If you're interested in hacking this possible future, let me know!
Just so you know how ridiculous
the current set up is.
Product vision journal ->
Hi friends, didn't do Pioneer update this week. Decided to move away from Pioneer to straightly focus on building instead + take care of other aspects of life, such as corona crisis world.
Software side -
Made great progress on computer vis. :)
Hi friends, welcome to another week of update. This week, we've made big progress on building out annotation use case. Exciting things coming.
Product side - Developing annotation use case, will show more later. For brief sketch, see about. Started a team, and mood board.
Hardware side - Considering new clamp for holding the device.
Software side - Explored web client! Also, made progress on object detection. Next step is to make it real time.
Hope you all have a great leap day!
I hear you on the feedback on being more clear in my progress update, so I decided to write out the details in paragraph.
Product side - answering some questions about "why do you do this project?", "what do you mean by bringing interfaces to books" at About page. TLDR: this is an exploratory project on improving my own reading experience, and if it works, maybe it could help improve others' experience too.
Hardware side - designed and 3D printed a case for the device! (HUGE thank you to Work hard, play hard@NB for helping me out) Why we need a case? Because it takes time to reposition the camera and projector - having a case will stabilizes the movement and enables me to go into software dev mode faster! This is 1st iteration. 2nd iteration is in the pipeline.
Software side - did a tutorial on simple shape recognition, and read a ton of documentation on the openCV methods. (see my learning log) Now I understood (in general sense) how shape recognition works -- next step is to apply my learning to the camera script, so the camera can recognize where the book is.
Drew out what information
will be projected on books
Top view of the device:
camera projector offset is important. That's why we need a case.
3D model of the case
Case refinement plan
Learning shape recognition algo
Projector/camera stable setup
Completed stable set up
Wrote basic camera script: read from desk
Wrote basic projector script: project frame
Got feedback on use cases - uncovered challenge: people don't read on desk
Started exploring input devices
Testing bounding box
Camera stream running
Projecting program from pi
Set up clamp and mounted projector
Created this website
Continue adding use cases for books
Booting up the pi
Setting up pi
with laser projector
Bookmark use case
Started last summer. Toyed with
a bunch of faulty projectors. Took me a
while to understand what needs to be done.
(turns out it's nothing new under the sun!)
"Sixth Sense on youtube, dope!"
- 2010 Tinnei's diary
will be added here
Feel free to check back if you're interested in following along my progress.