Robin Cole

Create class for ROI

2

The ROI should be abstracted using a class. The class will implement a `point_in_roi` method that accepts a single `x,y` location and returns a boolean. We can keep a list...

See https://github.com/robmarkcole/HASS-amazon-rekognition/issues/102

Allow list of ROI and give each a name

2

A single camera might cover, e.g. a garden and a driveway. We might want both as distinct ROI. Furthermore we might have several cameras watching a single ROI and want...

Currently they only appear if there is a detection of the target. Update so that even if there are no detections of the target it appears in the summary with...

as title

Document custom models

1

Add section. Show running with correct logging and link to model store. ``` ~ docker run -v /Users/robin/Data/pytorch-models:/modelstore/detection -e VISION-DETECTION=True -p 80:5000 deepquestai/deepstack:latest --------------------------------------- v1/vision/custom/fire --------------------------------------- v1/vision/custom/licence-plate --------------------------------------- v1/vision/custom/lock ---------------------------------------...

passing `min_confidence` of zero (0) will cause Deepstack to hang indefinitely

2

Parallels https://github.com/johnolafenwa/DeepStack/issues/58 Can implement workaround here, coerce min confidence to 1%

If using a custom model put this in the entity name

The nomenclature ROI = region of interest, is similar to the Zone concept in HA. Consider renaming ROI to camera_zone, however this might be a bit confusing due to the...

An IOU calc would give the % overlap of a detected object bounding box with the target area, which is a more flexible detection approach than specifying that the centre...