#include "STrack.h" /** * @brief Constructs an STrack object with initial bounding box and detection score. * * This constructor initializes a new track with a bounding box in top-left-width-height (tlwh) format and a detection confidence score. * Initializing properties like track ID, state, and frame information. The track starts as unactivated and in the "New" state, ready for further processing. * * @param tlwh_ A vector of four floats representing the bounding box in [top, left, width, height] format. * @param score The confidence score of the detection associated with this track, typically from a detector. */ STrack::STrack(vector tlwh_, float score) { _tlwh.resize(4); _tlwh.assign(tlwh_.begin(), tlwh_.end()); is_activated = false; track_id = 0; state = TrackState::New; tlwh.resize(4); tlbr.resize(4); static_tlwh(); static_tlbr(); frame_id = 0; tracklet_len = 0; this->score = score; start_frame = 0; } STrack::~STrack() { } /** * @brief Activates a track by initializing its Kalman filter and setting its vector state. * * This function activates a new track by assigning it a unique track ID, initializing the Kalman filter with the track's bounding box * in center-based coordinates (xyah where x&y are center of a bbox, a=w/h) and updating its state to "Tracked". * It is typically called when a new detection is confirmed as a track. The track's initial frame and other properties are also set. * * @param kalman_filter Reference to a KalmanFilter object used to initialize the track's state estimation. * @param frame_id The ID of the current frame in the video sequence, used to set the track's starting frame. */ void STrack::activate(byte_kalman::KalmanFilter &kalman_filter, int frame_id) { this->kalman_filter = kalman_filter; this->track_id = this->next_id(); vector _tlwh_tmp(4); _tlwh_tmp[0] = this->_tlwh[0]; _tlwh_tmp[1] = this->_tlwh[1]; _tlwh_tmp[2] = this->_tlwh[2]; _tlwh_tmp[3] = this->_tlwh[3]; vector xyah = tlwh_to_xyah(_tlwh_tmp); DETECTBOX xyah_box; xyah_box[0] = xyah[0]; xyah_box[1] = xyah[1]; xyah_box[2] = xyah[2]; xyah_box[3] = xyah[3]; auto mc = this->kalman_filter.initiate(xyah_box); this->mean = mc.first; this->covariance = mc.second; static_tlwh(); static_tlbr(); this->tracklet_len = 0; this->state = TrackState::Tracked; if (frame_id == 1) { this->is_activated = true; } //this->is_activated = true; this->frame_id = frame_id; this->start_frame = frame_id; } /** * @brief Re-activates an existing track with updated information from a new detection. * * This function updates an existing track's state using a new detection's bounding box and score. It updates the Kalman filter * with the new bounding box in center-based coordinates, resets the tracklet length, and sets the track's state to "Tracked". * Optionally, it assigns a new track ID if requested. This is used to revive a track that was temporarily lost or unmatched. * * @param new_track Reference to an STrack object containing the new detection's bounding box and score. * @param frame_id The ID of the current frame in the video sequence, used to update the track's frame information. * @param new_id Boolean indicating whether to assign a new track ID (true) or retain the existing one (false). */ void STrack::re_activate(STrack &new_track, int frame_id, bool new_id) { vector xyah = tlwh_to_xyah(new_track.tlwh); DETECTBOX xyah_box; xyah_box[0] = xyah[0]; xyah_box[1] = xyah[1]; xyah_box[2] = xyah[2]; xyah_box[3] = xyah[3]; auto mc = this->kalman_filter.update(this->mean, this->covariance, xyah_box); this->mean = mc.first; this->covariance = mc.second; static_tlwh(); static_tlbr(); this->tracklet_len = 0; this->state = TrackState::Tracked; this->is_activated = true; this->frame_id = frame_id; this->score = new_track.score; if (new_id) this->track_id = next_id(); } /** * @brief Updates an existing track with new detection information. * * This function updates the track's state using a new detection's bounding box and score. It increments the tracklet length, * updates the Kalman filter with the new bounding box in center-based coordinates (xyah), and sets the track's state to "Tracked" * and activated. This is used to refine a track's position and attributes when matched with a new detection. * * @param new_track Reference to an STrack object containing the new detection's bounding box and score. * @param frame_id The ID of the current frame in the video sequence, used to update the track's frame information. */ void STrack::update(STrack &new_track, int frame_id) { this->frame_id = frame_id; this->tracklet_len++; vector xyah = tlwh_to_xyah(new_track.tlwh); DETECTBOX xyah_box; xyah_box[0] = xyah[0]; xyah_box[1] = xyah[1]; xyah_box[2] = xyah[2]; xyah_box[3] = xyah[3]; auto mc = this->kalman_filter.update(this->mean, this->covariance, xyah_box); this->mean = mc.first; this->covariance = mc.second; static_tlwh(); static_tlbr(); this->state = TrackState::Tracked; this->is_activated = true; this->score = new_track.score; } /** * @brief Updates the track's bounding box in top-left-width-height (tlwh) format based on its state. * * This function sets the track's tlwh member (top-left-width-height bounding box) based on its current state. For new tracks, * it copies the initial _tlwh values directly. It converts the Kalman filter's mean state (assumed to be in center-based xyah format: * center-x, center-y, aspect ratio (width / height), height) to tlwh format. This is used to maintain consistent bounding box representation. */ void STrack::static_tlwh() { if (this->state == TrackState::New) { tlwh[0] = _tlwh[0]; tlwh[1] = _tlwh[1]; tlwh[2] = _tlwh[2]; tlwh[3] = _tlwh[3]; return; } tlwh[0] = mean[0]; tlwh[1] = mean[1]; tlwh[2] = mean[2]; tlwh[3] = mean[3]; tlwh[2] *= tlwh[3]; tlwh[0] -= tlwh[2] / 2; tlwh[1] -= tlwh[3] / 2; } /** * This function converts the track's bounding box from top-left-width-height (tlwh) format to top-left-bottom-right (tlbr) */ void STrack::static_tlbr() { tlbr.clear(); tlbr.assign(tlwh.begin(), tlwh.end()); tlbr[2] += tlbr[0]; tlbr[3] += tlbr[1]; } /** * @brief Converts a bounding box from top-left-width-height (tlwh) to center-x, center-y, aspect-ratio (w/h), height (xyah) format. * * This function transforms a bounding box from tlwh format [top, left, width, height] to xyah format [center_x, center_y, aspect_ratio, height]. * * @param tlwh_tmp A vector of four floats representing the bounding box in [top, left, width, height] format. * @return A vector of four floats representing the bounding box in [center_x, center_y, aspect_ratio, height] format. */ vector STrack::tlwh_to_xyah(vector tlwh_tmp) { vector tlwh_output = tlwh_tmp; tlwh_output[0] += tlwh_output[2] / 2; tlwh_output[1] += tlwh_output[3] / 2; tlwh_output[2] /= tlwh_output[3]; return tlwh_output; } vector STrack::to_xyah() { return tlwh_to_xyah(tlwh); } // convert vector state format from top-left-bottom-right to top-left-width-height vector STrack::tlbr_to_tlwh(vector &tlbr) { tlbr[2] -= tlbr[0]; tlbr[3] -= tlbr[1]; return tlbr; } // mark a TrackState to 'Lost' void STrack::mark_lost() { state = TrackState::Lost; } // mark a TrackState to 'Removed' void STrack::mark_removed() { state = TrackState::Removed; } // generate the next unique track ID for a new track. int STrack::next_id() { static int _count = 0; _count++; return _count; } // returns the most recent frame ID associated with the track. int STrack::end_frame() { return this->frame_id; } /** * @brief Performs motion prediction for multiple tracks using a Kalman filter and global motion compensation. * * This function updates the state (mean and covariance) of multiple tracks by applying a global motion transformation (camera motion compensation) * and then performing a Kalman filter prediction step. It is used to predict the next position * of tracks across frames, accounting for global motion. The bounding box representations (tlwh and tlbr) are updated accordingly. * * @param stracks Vector of pointers to STrack objects representing the tracks to be predicted. * @param kalman_filter Reference to a KalmanFilter object used for state prediction. * @param M Eigen matrix representing the global motion model, calculated in BYTETracker::update(...) */ void STrack::multi_predict(vector &stracks, byte_kalman::KalmanFilter &kalman_filter, Eigen::MatrixXf M) { Eigen::MatrixXf m_homo = M(Eigen::all, Eigen::seq(0, 7)); Eigen::VectorXf m_trans = M.col(8); for (int i = 0; i < stracks.size(); i++) { if (stracks[i]->state != TrackState::Tracked) { stracks[i]->mean[7] = 0; } stracks[i]->mean = stracks[i]->mean * m_homo.transpose() + m_trans.transpose(); stracks[i]->covariance = m_homo * stracks[i]->covariance * m_homo.transpose(); kalman_filter.predict(stracks[i]->mean, stracks[i]->covariance); stracks[i]->static_tlwh(); stracks[i]->static_tlbr(); } }