Shared workers seem to cause problems with iOS (and sharing wasn't really required anyway), so this PR changes the shared workers to non-shared workers. As a benefit, it preloads the full model and video models simultaneously which iproves performance when starting the video and running post video detection.
Signed-off-by: Justin Georgi <justin.georgi@gmail.com>
Reviewed-on: #191
The PR uses the vite recommend method of calling the shared worker so that PWA builds properly.
Signed-off-by: Justin Georgi <justin.georgi@gmail.com>
Reviewed-on: #188
This PR:
* Adds color coded markers on the confidence slider
* Changes the confidence slider visual to be more intuitive (the highlight region covers the visible structures)
* Gives the tag container a max height so that too many tags doesn't reduce the image to 0 size
Signed-off-by: Justin Georgi <justin.georgi@gmail.com>
Reviewed-on: #185
Closes: #151
Models and samples are now properly structured in the vite public folder so they are properly automatically incorporated into the the builds and work with both builds and dev tests.
Reviewed-on: #152
Closes: #30
When the camera is being used to find an image to capture, the region mini model now runs in real time to give an estimate of where there are identifiable structures.
Reviewed-on: #143
Closes: #117
This bumps the thorax model from the yolo nano to the yolo sm (64 x 640 size) but greatly improves model performance by running a fake detection event on page load to get the model parameters in memory.
As a result of that change a new loading message was required so the f7 preloader was switched out for an f7 progress bar and more messaging was added during various stages. The progress bar fixes the previous issue with the preloader.
Signed-off-by: Justin Georgi <justin.georgi@gmail.com>
Reviewed-on: #125
This adds direct javascript access to the camera (with permission requests). Capturing an image with the camera now happens right on the detect page and not a separate window.
This also opens the possibility of live detection on the web app.
Reviewed-on: Georgi_Lab/ALVINN_f7#116
This updates the local detection to yolov8-based models from the older yolov5 based models.
This includes significant additional post-processing of the raw yolov8 output, but also means that the smaller nano yolo model is available as well as all of the updated ALVINN thorax training data.
Signed-off-by: Justin Georgi <justin.georgi@gmail.com>
Reviewed-on: Georgi_Lab/ALVINN_f7#115
Tensorflow fetch failed on android with basic configuration. More specific protocol and location for model.json required.
Signed-off-by: Justin Georgi <justin.georgi@gmail.com>
Reviewed-on: Georgi_Lab/ALVINN_f7#99
