Example Devices & Latencies

The following table shows performance metrics for various devices for each of the three model options for AutoML Vision Edge.

These values should be used for guidance purposes only and not guarantees of performance; your latency values may vary. Actual latency will depend on your target device and environment setup.

This list is also not meant to serve as an exhaustive list of devices supported.

Device Model option Latency Size Accuracy
Google Pixel 1 Lowest latency 22 milliseconds 557 Kb Typically lower
General purpose 65 milliseconds 3.1 MB Best tradeoff
Higher accuracy 105 milliseconds 5.6 MB Typically higher
Google Pixel 2 XL Lowest latency 22 milliseconds 557 Kb Typically lower
General purpose 65 milliseconds 3.1 MB Best tradeoff
Higher accuracy 94 milliseconds 5.6 MB Typically higher
Samsung Galaxy S7 Lowest latency 22 milliseconds 557 Kb Typically lower
General purpose 66 milliseconds 3.1 MB Best tradeoff
Higher accuracy 102 milliseconds 5.6 MB Typically higher
Samsung Galaxy Note 3 Lowest latency 38 milliseconds 557 Kb Typically lower
General purpose 103 milliseconds 3.1 MB Best tradeoff
Higher accuracy 148 milliseconds 5.6 MB Typically higher
LG G6 Lowest latency 20 milliseconds 557 Kb Typically lower
General purpose 62 milliseconds 3.1 MB Best tradeoff
Higher accuracy 93 milliseconds 5.6 MB Typically higher
iPhone 8 + iOS11 Lowest latency 8 milliseconds 557 Kb Typically lower
General purpose 23 milliseconds 3.1 MB Best tradeoff
Higher accuracy 34 milliseconds 5.6 MB Typically higher
Edge TPU Lowest latency 2 milliseconds 858 Kb Typically lower
General purpose 3 milliseconds 3.7 MB Best tradeoff
Higher accuracy 5 milliseconds 6.8 MB Typically higher
Esta página foi útil? Conte sua opinião sobre:

Enviar comentários sobre…

Cloud AutoML Vision
Precisa de ajuda? Acesse nossa página de suporte.