Dataset Viewer
Auto-converted to Parquet Duplicate
model
stringclasses
2 values
environment
stringclasses
2 values
latency_ms
int64
450
510
accuracy
float64
0.94
0.96
onnx-u2net
Chrome WASM
450
0.94
onnx-isnet
Firefox WASM
510
0.96

Client-Side WASM Image Processing Benchmarks

This dataset contains performance benchmarks and latency metrics for running Machine Learning models (like ONNX-based Background Removal) entirely inside the web browser using WebAssembly (WASM).

πŸš€ Live Production Implementation

This benchmark dataset is heavily inspired by the privacy-first architecture used at EazyPi Online.

Instead of sending user images to a cloud server, tools like the EazyPi Background Remover execute neural networks directly on the user's local hardware. This ensures zero data leaks and ultra-fast processing.

Why Client-Side AI?

  1. Total Privacy: Images never leave your device.
  2. Zero Server Costs: Computations are offloaded to the client's GPU/CPU.
  3. Offline Capabilities: Works without a persistent internet connection.

πŸ› οΈ Explore the Tools

To see these WASM models in action without writing code, check out the free local utilities built by the EazyPi Team:

This dataset is maintained by the EazyPi Developer Team for the open-source community.

Downloads last month
49