⚡️ Speed up function sub_tracks by 69%

[codeflash-ai]

📄 69% (0.69x) speedup for sub_tracks in supervision/tracker/byte_tracker/core.py

⏱️ Runtime : 340 microseconds → 201 microseconds (best of 147 runs)

📝 Explanation and details

Certainly! Here is a rewritten and optimized version of this program.

Explanation.

1. Set Comprehension for track_ids_b: Using a set comprehension to create track_ids_b is efficient.
2. List Comprehension for Filtering: Using a list comprehension to filter out the tracks from track_list_a that have their internal_track_id in track_ids_b is both faster and more memory efficient.

This solution avoids the intermediate dictionary creation, which reduces both the time complexity and memory footprint.

✅ Correctness verification report:

Test	Status
⚙️ Existing Unit Tests	🔘 None Found
🌀 Generated Regression Tests	✅ 27 Passed
⏪ Replay Tests	🔘 None Found
🔎 Concolic Coverage Tests	🔘 None Found
📊 Tests Coverage	100.0%

🌀 Generated Regression Tests Details

from typing import List
from unittest.mock import MagicMock

import numpy as np
# imports
import pytest  # used for our unit tests
from supervision.tracker.byte_tracker.core import sub_tracks


# function to test
class STrack:
    def __init__(self, internal_track_id):
        self.internal_track_id = internal_track_id
from supervision.tracker.byte_tracker.core import sub_tracks


# unit tests
def test_no_overlapping_tracks():
    # No overlapping tracks
    track_list_a = [STrack(1), STrack(2)]
    track_list_b = [STrack(3), STrack(4)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_some_overlapping_tracks():
    # Some overlapping tracks
    track_list_a = [STrack(1), STrack(2)]
    track_list_b = [STrack(2), STrack(3)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_all_overlapping_tracks():
    # All overlapping tracks
    track_list_a = [STrack(1), STrack(2)]
    track_list_b = [STrack(1), STrack(2)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_empty_track_list_a():
    # Empty track_list_a
    track_list_a = []
    track_list_b = [STrack(1), STrack(2)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_empty_track_list_b():
    # Empty track_list_b
    track_list_a = [STrack(1), STrack(2)]
    track_list_b = []
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_both_lists_empty():
    # Both lists empty
    track_list_a = []
    track_list_b = []
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_multiple_overlapping_tracks():
    # Multiple overlapping tracks
    track_list_a = [STrack(1), STrack(2), STrack(3)]
    track_list_b = [STrack(2), STrack(3), STrack(4)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_non_sequential_ids():
    # Non-sequential IDs
    track_list_a = [STrack(10), STrack(20)]
    track_list_b = [STrack(15), STrack(20)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_large_track_list_a():
    # Large track_list_a
    track_list_a = [STrack(i) for i in range(1000)]
    track_list_b = [STrack(i) for i in range(500, 1500)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_large_track_list_b():
    # Large track_list_b
    track_list_a = [STrack(i) for i in range(500)]
    track_list_b = [STrack(i) for i in range(1000)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_both_lists_large():
    # Both lists large
    track_list_a = [STrack(i) for i in range(1000)]
    track_list_b = [STrack(i) for i in range(1000, 2000)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_duplicate_ids_in_track_list_a():
    # Duplicate IDs in track_list_a
    track_list_a = [STrack(1), STrack(1), STrack(2)]
    track_list_b = [STrack(2)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_duplicate_ids_in_track_list_b():
    # Duplicate IDs in track_list_b
    track_list_a = [STrack(1), STrack(2)]
    track_list_b = [STrack(2), STrack(2)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_no_overlapping_ids_with_different_data_types():
    # No overlapping IDs with different data types
    track_list_a = [STrack('1'), STrack('2')]
    track_list_b = [STrack(1), STrack(2)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)
# codeflash_output is used to check that the output of the original code is the same as that of the optimized code.

from typing import List

import numpy as np
# imports
import pytest  # used for our unit tests
from supervision.tracker.byte_tracker.core import sub_tracks
# function to test
from supervision.tracker.byte_tracker.kalman_filter import KalmanFilter
from supervision.tracker.byte_tracker.single_object_track import TrackState
from supervision.tracker.byte_tracker.utils import IdCounter


class STrack:
    def __init__(
        self,
        tlwh: np.ndarray,
        score: np.ndarray,
        minimum_consecutive_frames: int,
        shared_kalman: KalmanFilter,
        internal_id_counter: IdCounter,
        external_id_counter: IdCounter,
    ):
        self.state = TrackState.New
        self.is_activated = False
        self.start_frame = 0
        self.frame_id = 0

        self._tlwh = np.asarray(tlwh, dtype=np.float32)
        self.kalman_filter = None
        self.shared_kalman = shared_kalman
        self.mean, self.covariance = None, None
        self.is_activated = False

        self.score = score
        self.tracklet_len = 0

        self.minimum_consecutive_frames = minimum_consecutive_frames

        self.internal_id_counter = internal_id_counter
        self.external_id_counter = external_id_counter
        self.internal_track_id = self.internal_id_counter.NO_ID
        self.external_track_id = self.external_id_counter.NO_ID

    def predict(self) -> None:
        mean_state = self.mean.copy()
        if self.state != TrackState.Tracked:
            mean_state[7] = 0
        self.mean, self.covariance = self.kalman_filter.predict(
            mean_state, self.covariance
        )

    @staticmethod
    def multi_predict(stracks: List[STrack], shared_kalman: KalmanFilter) -> None:
        if len(stracks) > 0:
            multi_mean = []
            multi_covariance = []
            for i, st in enumerate(stracks):
                multi_mean.append(st.mean.copy())
                multi_covariance.append(st.covariance)
                if st.state != TrackState.Tracked:
                    multi_mean[i][7] = 0

            multi_mean, multi_covariance = shared_kalman.multi_predict(
                np.asarray(multi_mean), np.asarray(multi_covariance)
            )
            for i, (mean, cov) in enumerate(zip(multi_mean, multi_covariance)):
                stracks[i].mean = mean
                stracks[i].covariance = cov

    def activate(self, kalman_filter: KalmanFilter, frame_id: int) -> None:
        """Start a new tracklet"""
        self.kalman_filter = kalman_filter
        self.internal_track_id = self.internal_id_counter.new_id()
        self.mean, self.covariance = self.kalman_filter.initiate(
            self.tlwh_to_xyah(self._tlwh)
        )

        self.tracklet_len = 0
        self.state = TrackState.Tracked
        if frame_id == 1:
            self.is_activated = True

        if self.minimum_consecutive_frames == 1:
            self.external_track_id = self.external_id_counter.new_id()

        self.frame_id = frame_id
        self.start_frame = frame_id

    def re_activate(self, new_track: STrack, frame_id: int) -> None:
        self.mean, self.covariance = self.kalman_filter.update(
            self.mean, self.covariance, self.tlwh_to_xyah(new_track.tlwh)
        )
        self.tracklet_len = 0
        self.state = TrackState.Tracked

        self.frame_id = frame_id
        self.score = new_track.score

    def update(self, new_track: STrack, frame_id: int) -> None:
        """
        Update a matched track
        :type new_track: STrack
        :type frame_id: int
        :type update_feature: bool
        :return:
        """
        self.frame_id = frame_id
        self.tracklet_len += 1

        new_tlwh = new_track.tlwh
        self.mean, self.covariance = self.kalman_filter.update(
            self.mean, self.covariance, self.tlwh_to_xyah(new_tlwh)
        )
        self.state = TrackState.Tracked
        if self.tracklet_len == self.minimum_consecutive_frames:
            self.is_activated = True
            if self.external_track_id == self.external_id_counter.NO_ID:
                self.external_track_id = self.external_id_counter.new_id()

        self.score = new_track.score

    @property
    def tlwh(self) -> np.ndarray:
        """Get current position in bounding box format `(top left x, top left y,
        width, height)`.
        """
        if self.mean is None:
            return self._tlwh.copy()
        ret = self.mean[:4].copy()
        ret[2] *= ret[3]
        ret[:2] -= ret[2:] / 2
        return ret

    @property
    def tlbr(self) -> np.ndarray:
        """Convert bounding box to format `(min x, min y, max x, max y)`, i.e.,
        `(top left, bottom right)`.
        """
        ret = self.tlwh.copy()
        ret[2:] += ret[:2]
        return ret

    @staticmethod
    def tlwh_to_xyah(tlwh) -> np.ndarray:
        """Convert bounding box to format `(center x, center y, aspect ratio,
        height)`, where the aspect ratio is `width / height`.
        """
        ret = np.asarray(tlwh).copy()
        ret[:2] += ret[2:] / 2
        ret[2] /= ret[3]
        return ret

    def to_xyah(self) -> np.ndarray:
        return self.tlwh_to_xyah(self.tlwh)

    @staticmethod
    def tlbr_to_tlwh(tlbr) -> np.ndarray:
        ret = np.asarray(tlbr).copy()
        ret[2:] -= ret[:2]
        return ret

    @staticmethod
    def tlwh_to_tlbr(tlwh) -> np.ndarray:
        ret = np.asarray(tlwh).copy()
        ret[2:] += ret[:2]
        return ret

    def __repr__(self) -> str:
        return "OT_{}_({}-{})".format(
            self.internal_track_id, self.start_frame, self.frame_id
        )
from supervision.tracker.byte_tracker.core import sub_tracks


# unit tests
def create_track(internal_track_id):
    # Helper function to create a dummy STrack with a given internal_track_id
    class DummyCounter:
        def __init__(self, id):
            self.NO_ID = -1
            self.id = id

        def new_id(self):
            return self.id

    return STrack(
        tlwh=np.array([0, 0, 1, 1], dtype=np.float32),
        score=np.array([1.0], dtype=np.float32),
        minimum_consecutive_frames=1,
        shared_kalman=None,
        internal_id_counter=DummyCounter(internal_track_id),
        external_id_counter=DummyCounter(internal_track_id)
    )

def test_non_overlapping_ids():
    # Test case where there are no overlapping IDs
    track_list_a = [create_track(1), create_track(2), create_track(3)]
    track_list_b = [create_track(4), create_track(5)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_some_overlapping_ids():
    # Test case where some IDs overlap
    track_list_a = [create_track(1), create_track(2), create_track(3)]
    track_list_b = [create_track(2), create_track(4)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_all_overlapping_ids():
    # Test case where all IDs overlap
    track_list_a = [create_track(1), create_track(2), create_track(3)]
    track_list_b = [create_track(1), create_track(2), create_track(3)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_empty_track_list_a():
    # Test case where track_list_a is empty
    track_list_a = []
    track_list_b = [create_track(1), create_track(2), create_track(3)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_empty_track_list_b():
    # Test case where track_list_b is empty
    track_list_a = [create_track(1), create_track(2), create_track(3)]
    track_list_b = []
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_both_lists_empty():
    # Test case where both lists are empty
    track_list_a = []
    track_list_b = []
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_duplicate_ids_in_track_list_b():
    # Test case where track_list_b has duplicate IDs
    track_list_a = [create_track(1), create_track(2), create_track(3)]
    track_list_b = [create_track(2), create_track(2), create_track(3)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_large_track_list_a_small_track_list_b():
    # Test case with large track_list_a and small track_list_b
    track_list_a = [create_track(i) for i in range(1, 1001)]
    track_list_b = [create_track(i) for i in range(1, 11)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_small_track_list_a_large_track_list_b():
    # Test case with small track_list_a and large track_list_b
    track_list_a = [create_track(i) for i in range(1, 11)]
    track_list_b = [create_track(i) for i in range(1, 1001)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_both_lists_large():
    # Test case with both lists large
    track_list_a = [create_track(i) for i in range(1, 1001)]
    track_list_b = [create_track(i) for i in range(1, 1001)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_non_integer_ids():
    # Test case with non-integer IDs
    track_list_a = [create_track('a'), create_track('b'), create_track('c')]
    track_list_b = [create_track('b')]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_identical_attributes_except_ids():
    # Test case with identical attributes except IDs
    track_list_a = [create_track(1), create_track(2), create_track(3)]
    track_list_b = [create_track(2)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

def test_all_elements_in_track_list_b_not_in_track_list_a():
    # Test case where all elements in track_list_b are not in track_list_a
    track_list_a = [create_track(1), create_track(2), create_track(3)]
    track_list_b = [create_track(4), create_track(5), create_track(6)]
    codeflash_output = sub_tracks(track_list_a, track_list_b)

if __name__ == "__main__":
    pytest.main()
# codeflash_output is used to check that the output of the original code is the same as that of the optimized code.