Minimum Hardware Specification for SubCenter-ArcFace-r100

**I am using the SubCenter-ArcFace-r100 custom build for face verification. My server config is as follows:**

1. cpu : Intel(R) Xeon(R) CPU E5-2699 v4 @ 2.20GHz, using 12 cores.
2. ram : 16 gb
3. This is a virtual machine. running windows 10 pro OS.
4. Containers running on docker desktop 4.27.2
Everything working fine and the accuracy is good. But the problem i am facing is the response time of an api request is almost **22 seconds**. How can I **reduce this time to 5 to 10 seconds** at least?
**My expectation : response within 5-10 seconds.
what would be the hardware and software requirements?**