Hardware Efficiency and Performance on Medical Image Classification on MedMNIST
192.3Throughput (Inference/Sec)MedMambaLite-ST
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| MedMambaLite-STPlatform=Lambda Server2025.08 | 192.3 | — | 29.2 | — | — | |
| MedMambaLite-TRPlatform=Lambda Server2025.08 | 185.2 | — | 86.9 | — | — | |
| MedMamba-TPlatform=Lambda Server2025.08 | 149.3 | — | 338.1 | — | — | |
| MedMambaLite-STPlatform=NVIDIA Jetson Orin Nano with 8 GB Memory2025.08 | 76.7 | 2.7 | 11.7 | 327.6 | 35.6 | |
| MedMambaLite-TRPlatform=NVIDIA Jetson Orin Nano with 8 GB Memory2025.08 | 63.7 | 3.3 | 29.9 | 571.2 | 52.3 | |
| MedMambaPlatform=NVIDIA Jetson Orin Nano with 8 GB Memory2025.08 | 34.2 | 3.2 | 77.4 | 814.5 | 95 | |
| MedMambaLite-STPlatform=Raspberry Pi 5 with 16 GB Memory2025.08 | 18.9 | 7.7 | 2.9 | 7.1 | 405.6 | |
| MedMambaLite-TRPlatform=Raspberry Pi 5 with 16 GB Memory2025.08 | 9.1 | 6.1 | 4.2 | 6.3 | 678.5 | |
| MedMambaPlatform=Raspberry Pi 5 with 16 GB Memory2025.08 | 4.3 | 6.7 | 9.8 | 6.3 | 1,560 |