Regarding the issue of WASM operating speed

When using the YOLO project in the handle wasm example, it takes a long time to make predictions on the browser side, and the CPU usage remains unchanged. How can I improve prediction performance.

The issue points to the YOLO project in the handle wasm example, but names no file or test. Start by reproducing browser-side prediction and measuring runtime and CPU usage, then trace the WASM inference path to identify the bottleneck. Done means prediction performance is measurably improved and the result is verified in the browser example.

Written by the indexing model from the issue text.