Method update

tutorials/cstrack/tracker.py:207–379

(self, im_blob, img0,seq_num, save_dir)

        self.high_thres = self.opt.conf_thres + 0.1

    def update(self, im_blob, img0,seq_num, save_dir):
        self.frame_id += 1
        activated_starcks = []
        refind_stracks = []
        lost_stracks = []
        removed_stracks = []
        dets = []

        ''' Step 1: Network forward, get detections & embeddings'''
        with torch.no_grad():
            output = self.model(im_blob, augment=False)
            pred, train_out = output[1]

        pred = pred[pred[:, :, 4] > self.low_thres]
        detections = []
        if len(pred) > 0:
            dets,x_inds,y_inds = non_max_suppression_and_inds(pred[:,:6].unsqueeze(0), 0.1, self.opt.nms_thres,method='cluster_diou')
            dets = dets.numpy()
            if len(dets) != 0:
                scale_coords(self.opt.img_size, dets[:, :4], img0.shape).round()
                id_feature = output[0][0, y_inds, x_inds, :].cpu().numpy()

                remain_inds = dets[:, 4] > self.opt.conf_thres
                inds_low = dets[:, 4] > self.low_thres
                inds_high = dets[:, 4] < self.opt.conf_thres
                inds_second = np.logical_and(inds_low, inds_high)
                dets_second = dets[inds_second]
                if id_feature.shape[0] == 1:
                    id_feature_second = id_feature
                else:
                    id_feature_second = id_feature[inds_second]
                dets = dets[remain_inds]
                id_feature = id_feature[remain_inds]

                detections = [STrack(STrack.tlbr_to_tlwh(tlbrs[:4]), tlbrs[4], f, 30) for
                              (tlbrs, f) in zip(dets[:, :5], id_feature)]
                
            else:
                detections = []
                dets_second = []
                id_feature_second = []
        else:
            detections = []
            dets_second = []
            id_feature_second = []

        ''' Add newly detected tracklets to tracked_stracks'''
        unconfirmed = []
        tracked_stracks = []  # type: list[STrack]
        for track in self.tracked_stracks:
            if not track.is_activated:
                unconfirmed.append(track)
            else:
                tracked_stracks.append(track)

        ''' Step 2: First association, with embedding'''
        strack_pool = joint_stracks(tracked_stracks, self.lost_stracks)
        # Predict the current location with KF