phil-bergmann / tracking_wo_bnw

Implementation of "Tracking without bells and whistles” and the multi-object tracking "Tracktor"

Home Page:https://arxiv.org/abs/1903.05625

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Tracks found: 0(python version:3.7.7, torch:1.3.0, torchvision: 0.4.1)

chenrxi opened this issue · comments

commented

WARNING - test_tracktor - No observers have been added to this run
INFO - test_tracktor - Running command 'main'
INFO - test_tracktor - Started
Configuration (modified, added, typechanged, doc):
seed = 453695
reid:
db_train = 'train'
db_val = False
desription = 'all'
lr_scheduler = True
max_epochs = 69
module_name = 'reid'
name = 'res50-mot17-batch_hard'
seed = 12345
cnn:
output_dim = 128
dataloader:
K = 4
P = 18
crop_H = 256
crop_W = 128
max_per_person = 1000
normalize_mean = [0.485, 0.456, 0.406]
normalize_std = [0.229, 0.224, 0.225]
transform = 'random'
vis_threshold = 0.3
model_args:
loss = 'batch_hard'
margin = 0.2
prec_at_k = 3
solver:
optim = 'Adam'
optim_args:
lr = 0.0003
weight_decay = 0.0
tracktor:
dataset = 'mot17_train_FRCNN17'
desription = None
frame_split = [0.0, 1.0]
interpolate = False
module_name = 'MOT17'
name = 'Tracktor++'
network = 'fpn'
obj_detect_model = 'output/faster_rcnn_fpn_training_mot_17/model_epoch_27.model'
reid_config = 'output/tracktor/reid/res50-mot17-batch_hard/sacred_config.yaml'
reid_weights = 'output/tracktor/reid/res50-mot17-batch_hard/ResNet_iter_25245.pth'
seed = 12345
write_images = False
tracker:
detection_nms_thresh = 0.3
detection_person_thresh = 0.5
do_align = True
do_reid = True
inactive_patience = 10
max_features_num = 10
number_of_iterations = 100
public_detections = True
regression_nms_thresh = 0.6
regression_person_thresh = 0.5
reid_iou_threshold = 0.2
reid_sim_threshold = 2.0
termination_eps = 1e-05
warp_mode = 'cv2.MOTION_EUCLIDEAN'
motion_model:
center_only = True
enabled = False
n_steps = 1
INFO - main - Initializing object detector.
INFO - main - Tracking: MOT17-02-FRCNN
100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 600/600 [01:15<00:00, 7.97it/s]
INFO - main - Tracks found: 0
INFO - main - Runtime for MOT17-02-FRCNN: 75.33 s.
INFO - main - Writing predictions to: /home/chenruoxi/tracking_wo_bnw/output/tracktor/MOT17/Tracktor++
INFO - main - Tracking: MOT17-04-FRCNN

What can i do?

Hello, please double check if the dataset is in the correct directory and if you completed all other steps of the README.md.

I found that:
public_detections = True
was searching for external detections which could not be found in my case, then it found 0 tracks.

Set it to False, solved my issue.

I suspect a dataset issue in both cases. The MOT17 dataset folder includes public detection sets.

Hmm, true. I thought it was a custom dataset, I was a bit too quick in my reply :)