Efficient Inference on Emergency response dataset 2500 images (test)
657FPSTakuNet
Evaluation results
| Method | Links | |
|---|---|---|
| TakuNetDevice=Jetson Orin Nano, Accelerator=GPU Tegra (Ampere), TDP=15W, Batch size=1, Image size=240 pixels, Optimization=TensorRT2025.01 | 657 | |
| TakuNetDevice=Jetson Orin Nano, Accelerator=GPU Tegra (Ampere), TDP=7W, Batch size=1, Image size=240 pixels, Optimization=TensorRT2025.01 | 564 | |
| TinyEmergencyNetDevice=Jetson Orin Nano, Accelerator=GPU Tegra (Ampere), TDP=15W, Batch size=1, Image size=240 pixels2025.01 | 160 | |
| EmergencyNetDevice=Jetson Orin Nano, Accelerator=GPU Tegra (Ampere), TDP=15W, Batch size=1, Image size=240 pixels2025.01 | 140 | |
| TakuNetDevice=Raspberry Pi 5, Accelerator=CPU ARM Cortex-A76, TDP=10W, Batch size=1, Image size=240 pixels2025.01 | 62.1 | |
| TinyEmergencyNetDevice=Raspberry Pi 5, Accelerator=CPU ARM Cortex-A76, TDP=10W, Batch size=1, Image size=240 pixels2025.01 | 59.8 | |
| EmergencyNetDevice=Raspberry Pi 5, Accelerator=CPU ARM Cortex-A76, TDP=10W, Batch size=1, Image size=240 pixels2025.01 | 43.3 | |
| TinyEmergencyNetDevice=Raspberry Pi 4, Accelerator=CPU ARM Cortex-A72, TDP=10W, Batch size=1, Image size=240 pixels2025.01 | 22 | |
| TakuNetDevice=Raspberry Pi 4, Accelerator=CPU ARM Cortex-A72, TDP=10W, Batch size=1, Image size=240 pixels2025.01 | 16.7 | |
| EmergencyNetDevice=Raspberry Pi 4, Accelerator=CPU ARM Cortex-A72, TDP=10W, Batch size=1, Image size=240 pixels2025.01 | 14.7 |