Camera Resolution cause delay

[Vina_Alvionita]

I build a social distance monitoring project. i use person detection retail model to detect person and calculate the distance. the camera used has several options of resolution. when the camera resolution is set to 1280*960 (1.3M) or 1280*720(720P), the delay in the output video is less than 1 second. but when the camera resolution increased to 920*1080, the resulting delay is more than 5 seconds. why it was happen? is there a way to solved the delay problem while used high resolution camera?

CPU : Intel Core i5-10500

camera : 5MP

[Victor_G_Intel]

Hello Vina_Alvionita,

Thank you for posting on the Intel® communities.

To better assist you please provide the following report and information to continue with your support.

Intel® System Support Utility (Intel® SSU)

Download link: https://downloadcenter.intel.com/download/25293/Intel-System-Support-Utility-for-Windows-

• Open the application and click on "Scan" to see the system and device information
• By default, Intel® SSU will take you to the "Summary View".  
• Click on the menu where it says "Summary" to change to "Detailed View".  
• To save your scan, click on "Next", then "Save".

- When you say that you use “person detection retail model” you mean the Open VINO™ Toolkit?

I look forward to hearing from you.

Regards,

Victor G.

Intel Technical Support Technician 

[Vina_Alvionita]

Yes,  i use person detection retail model from Open VINO™ Toolkit. and here i attached the summary report 

[Victor_G_Intel]

Hello Vina_Alvionita,

Thank you for posting on the Intel® communities.

Your query will be best answered by our Open VINO support team. We will help you to move this post to the designated team to further assist you.

Best regards,

Victor G

Intel Technical Support Technician

[Peh_Intel]

Hi Vina,

Greetings to you.

Higher resolution required higher processing power. The delay issue happened mainly because your CPU would have been a bottleneck when running the inference engine at 1080p.

Perhaps you can use person-detection-retail-0013 model (which the model is smaller in size) instead of person-detection-retail-0002 model and also try on different precision of the model (FP32, FP16-INT8).

Regards,

Peh

[Peh_Intel]

Hi Vina,

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

Regards,

Peh