A keyboard controlled multitab browser that overlays OpenMMLab model processing result on the web pages open in it
- SwiftUI
- Unity
- https://github.com/open-mmlab/mmediting
- https://github.com/open-mmlab/mmflow
- https://github.com/open-mmlab/mmdetection3d
- https://github.com/open-mmlab/mmocr
- https://github.com/open-mmlab/mmaction2
- https://github.com/open-mmlab/mmrotate
Got builtin HTTP server/React native web interface for control of navigation, tab switching, audio selection
Receives keyboard/mouse input via web socket interface
A hot key for sending a tab to another device
Has built in ROS node: camera, gps movement, actuators and camera for browser navigation
- SwiftUI interface
- CoreML models - exported by MMDeploy to ONNX Runtime
- On device labeling generates a request to upload a screenshot of page to the learning service server
- streams out tabs as rtmp
- able to show the overlayed labels from remote server
-
hotkey:
1
..9
,0
-
Shows a webpage
-
Has address bar
- If search term is entered, google search is performed
-
button to enter Labeling view
-
any object that is labeled by user and recognized is highlighted. It can be interacted with Labeled object popup
- Allows to label an underlying webpage in a modal overlay
- Boxes
- Automatic segmentation
- Semantic autolabeling
- People
- Vehicles
- Hands
- People skeletons
- Objects recognizable by image nets
- UI elements
- checkbox Show object on composite view
- textfield Fully qualified a Semantic Path label for the object
- hotkey:
~
Shows all objects, whose fully qualified Semantic
- url
- url features
- title
- title features
- webpage features
- UI
- Video Player
- Button
- Text Field
- Text
- Real World Object
- Humanoid
- Human
- Bot
- Vehicle
- Truck
- Tank
- Car
- Train
- Boat
- Airplane
- Item
- Landscape
- Building
- Humanoid
#AR view
SwiftUI app that streams the contents of its WebViews to the Python server over RTMP
Python server runs several models on the video, serves results to the subscribers over MixtreamProtocol