Process detail of multi camera multi person python demo ver2020.3

[Lenard-Lim]

Hi,

May I know the detail process of the person re-identification part of the demo?

My question is:

If the camera 1 and camera 2 detect the object 1 together and camera 1 assign ID1 to object 1. Then when camera 2 detect object1.

1)Will it assign the new ID with bounding box to object 1 first then compare with camera 1 to check the identity in camera 2?

or

2) camera 2 can can analyze the object 1 and straight away assign ID 1  to object 1 in camera 2.

because these is in the overlapping condition. May I know which number will be process?

other question is for the quality measuring in the multi camera demo webpage. May I know 

file.json is which file? and line 3 to line 4 is it direct put the code inside?

python run_evaluate.py \

--history_file path/to/history/file.json \

--gt_files \

path/to/ground/truth/annotation/for/source_0.xml \

path/to/ground/truth/annotation/for/source_1.xml \

[Zulkifli_Intel]

Hello Lenard Lim,

Thank you for contacting us.

When camera 1 detects object 1, it will assign as ID0. When camera 2 detects object 1, it will make a comparison with the assigned ID in camera 1. If matched, it will assign object 1 as ID0. The reidentification is also a difficult task because the appearance features of objects may change dramatically on account of varied illuminations and viewing angles in two different cameras.   

--history_file path/to/history/file.json is the path to file in JSON format to save the result of the demo. You can create it by using the command line argument and specify the path that you want to save the file.json.

You can refer here for more information.

Regards,

Zulkifli

[Lenard-Lim]

Hi Zulkifli,

So could I say that when the camera 1 detect and assign the ID0 to object1 then camera 2 will do the comparison and assign ID0 to object 1 if the similarity of appearance features of object is higher. If the similarity is lower but still higher than difference. The camera will assign the another Id to the object1 first and need more processing time to analyze the object then reassign back the correct ID.

Regards,

Lenard Lim

[Zulkifli_Intel · Accepted solution]

Hello Allen Seng Hung Wong,

When camera 2 detects object 1, usually it will assign object 1 as a different ID than camera 1 due to various accounts such as illuminations and viewing angle in two different cameras. If the characteristic is similar, it will reassign object 1 as ID0.

Regards,

Zulkifli

[Zulkifli_Intel]

Hello Lenard Lim,

This thread will no longer be monitored since this issue has been resolved. If you need any additional information from Intel, please submit a new question.

Regards,

Zulkifli