Intel® Distribution of OpenVINO™ Toolkit
Community assistance about the Intel® Distribution of OpenVINO™ toolkit, OpenCV, and all aspects of computer vision-related on Intel® platforms.
6370 Discussions

Multi-target multi-camera person.py setting documentation

Duggy
New Contributor I
455 Views

Hi,

 

Is there any information/documentation regarding the settings in person.py? 

For example the settings:

obj_segm = namespace(
trg_classes=(1,)
)

mct_config = namespace(
time_window=20,
global_match_thresh=0.2,
bbox_min_aspect_ratio=0.2 #1.2
)

sct_config = namespace(
time_window=1, #10,
continue_time_thresh=2,
track_clear_thresh=5, #3000,
track_clear_thresh_reid=1000000, #3000,
ignore_track_clear_thresh=10,
match_threshold=0.25, #0.25
merge_thresh=0.15, #0.15,
n_clusters=4,
max_bbox_velocity=2.0, ##0.2,
detection_occlusion_thresh=0.7,
track_detection_iou_thresh=0.5, #0.5,
process_curr_features_number=0,
interpolate_time_thresh=10,
detection_filter_speed=0.6,
rectify_thresh=0.1
)

 

If we want to decrease the threshold of person detection, i.e. doesn't have to be set to high for a given location, but needs to be faster as apposed to another location where needs to be more accurate and speed is less of a factor. 

Therefore we need to tweak some of the settings above (in the person.py file) for example. 

However these settings are used to call the object detection engine and then go into a "black hole" of code that cant be followed. 

The question is, is there any documentation on each one of these settings so we know what to tweak to achieve a goal?

 

Much appreciated.

Labels (3)
0 Kudos
2 Replies
IntelSupport
Community Manager
428 Views

Hi Duggy,

 

Thanks for reaching out.

 

All information for Multi Camera Multi Target Python Demo is as available on the online documentation. There is no other information regarding the configurations parameters setting for the demo.

 

Meanwhile, I have tried to edit the config file and observed the output as below setting:

ct_config = namespace(

   time_window=5,

   continue_time_thresh=1,

   track_clear_thresh=1000,

   match_threshold=0.25,

   merge_thresh=0.15,

   n_clusters=10,

   max_bbox_velocity=0.01,

   detection_occlusion_thresh=0.07,

   track_detection_iou_thresh=0.05,

   process_curr_features_number=0,

   interpolate_time_thresh=5,

   detection_filter_speed=0.6,

   rectify_thresh=0.01

)

 

multi_camera_multi_tracking_LI.jpg

 

However, there is not much difference in output result when I change the selected parameters.

You might have to experiment with the parameters based on your application requirement.

 

Regards,

Aznie

 

0 Kudos
IntelSupport
Community Manager
405 Views

Hi Duggy,


This thread will no longer be monitored since we have provided information. If you need any additional information from Intel, please submit a new question.



Regards,

Aznie


0 Kudos
Reply