GPU memory consumption

Hi to everyone, I'm testing your library. I'm really interested to estimate the facial landmarks. I have a computer with a GPU, and I'm using the **blazeface** detector, which is the faster as it comments in the documentation. **After several tests, the memory is increasing until ~6.5 GB of GPU when I want to extract 2D landmarks and ~8.5 to extract 3D landmarks**. I've noticed as well the **size network** used is always "Large" and there is some other 3 types uncommented, so I'm wondering about the relation of it with GPU performance and consuming because I need to decrease in some way the amount of memory.

Thanks in advance.

The issue mentions using the blazeface detector and the 'Large' network size. Start by examining the model loading and inference code in the face-alignment library, particularly around GPU memory management. Look for configuration options for network size and detector settings. Profile memory usage during landmark extraction to identify leaks or inefficient allocations. The goal is to understand why memory grows to 6.5-8.5 GB and if switching network sizes affects this.

Scritto dal modello di indicizzazione a partire dal testo della issue.