2. Build neural network inference library

At the time of writing there are three back-ends available to run the neural network inference with:

Back-end	GPU Support	CPU Support	Inference Speed	Effort to install
TensorRT (default)	✔️	❌	🔥🔥🔥🔥	⚠️⚠️
OpenVino	(:heavy_check_mark:) not tested yet	✔️	🔥🔥🔥	⚠️
MXNet	✔️	✔️	🔥🔥	⚠️⚠️⚠️ (CPU only) / ⚠️⚠️⚠️⚠️ (GPU)
Torch	✔️	✔️	🔥	⚠️

---

Next part:

• 3. Build CrazyAra binary