[LineZone] - allow per class counting

[SkalskiP]

Description

Currently, sv.LineZone provides only aggregated counts - all classes are thrown into one bucket. In the past, many users have asked us to provide more granular - per class count. This can be achieved by adding class_in_count and class_out_count dictionaries that will store per-class counts.

API

class LineZone:

    def __init__(
        self,
        start: Point,
        end: Point,
        triggering_anchors: Iterable[Position] = (
            Position.TOP_LEFT,
            Position.TOP_RIGHT,
            Position.BOTTOM_LEFT,
            Position.BOTTOM_RIGHT,
        ),
    ):
        # Existing initialization code...

        self.class_in_count: Dict[int, int] = {}
        self.class_out_count: Dict[int, int] = {}

    def trigger(self, detections: Detections) -> Tuple[np.ndarray, np.ndarray]:
        crossed_in = np.full(len(detections), False)
        crossed_out = np.full(len(detections), False)

        # Required logic changes...

Additional

• Note: Please share a Google Colab with minimal code to test the new feature. We know it's additional work, but it will definitely speed up the review process. Each change must be tested by the reviewer. Setting up a local environment to do this is time-consuming. Please ensure that Google Colab can be accessed without any issues (make it public). Thank you! 🙏🏻

[RaghavvGupta]

Hey @SkalskiP, I was wondering to work on this. I have this logic in mind, pls confirm if this approach is what you guys want?

Dictionaries to keep track of counts for each class

self.class_in_count: Dict[str, int] = {} self.class_out_count: Dict[str, int] = {}

Updating per class-in and class-out count

class_label = detections.class_labels[i]
                if class_label not in self.class_in_count:
                    self.class_in_count[class_label] = 0
                self.class_in_count[class_label] += 1
            else:
                self.out_count += 1
                crossed_out[i] = True

                # Updating per-class out count
                class_label = detections.class_labels[i]
                if class_label not in self.class_out_count:
                    self.class_out_count[class_label] = 0
                self.class_out_count[class_label] += 1

[RaghavvGupta]

Hey @SkalskiP 😀👋 https://colab.research.google.com/drive/1HxRcfzqnS6axAlYPiPpe8WLblfdn6fC_?usp=sharing

This is what I was talking about. Please tell me whether this is what you want.

[SkalskiP]

Hi, @RaghavvGupta, Would that be the implementation of a current count or the total count?

[RaghavvGupta]

I have tried implementing per-class counting. It includes dictionaries class_in_count and class_out_count to store counts for each class separately. The counts within these dictionaries are updated during each call to the trigger method, representing the current count of objects for each class based on their movement across the line.

[SkalskiP]

@RaghavvGupta, submit PR so we can move forward with the review process.

[caiquejjx]

Hello, is this being worked on?

[LinasKo]

Hi @caiquejjx 👋

Yes, it is. We have an outstanding PR that we need to review.