FCN-ResNet50: Optimized for Qualcomm Devices

FCN_ResNet50 is a machine learning model that can segment images from the COCO dataset. It uses ResNet50 as a backbone.

This is based on the implementation of FCN-ResNet50 found here. This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the Qualcomm® AI Hub Models library to export with custom configurations. More details on model performance across various devices, can be found here.

Qualcomm AI Hub Models uses Qualcomm AI Hub Workbench to compile, profile, and evaluate this model. Sign up to run these models on a hosted Qualcomm® device.

Getting Started

There are two ways to deploy this model on your device:

Option 1: Download Pre-Exported Models

Below are pre-exported model assets ready for deployment.

Runtime	Precision	Chipset	SDK Versions	Download
ONNX	float	Universal	QAIRT 2.45, ONNX Runtime 1.25.0	Download
ONNX	w8a8	Universal	QAIRT 2.45, ONNX Runtime 1.25.0	Download
QNN_DLC	float	Universal	QAIRT 2.45	Download
QNN_DLC	w8a8	Universal	QAIRT 2.45	Download
TFLITE	float	Universal	QAIRT 2.45	Download
TFLITE	w8a8	Universal	QAIRT 2.45	Download

For more device-specific assets and performance metrics, visit FCN-ResNet50 on Qualcomm® AI Hub.

Option 2: Export with Custom Configurations

Use the Qualcomm® AI Hub Models Python library to compile and export the model with your own:

• Custom weights (e.g., fine-tuned checkpoints)
• Custom input shapes
• Target device and runtime configurations

This option is ideal if you need to customize the model beyond the default configuration provided here.

See our repository for FCN-ResNet50 on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.semantic_segmentation

Model Stats:

• Model checkpoint: COCO_WITH_VOC_LABELS_V1
• Input resolution: 224x224
• Number of output classes: 21
• Number of parameters: 33.0M
• Model size (float): 126 MB
• Model size (w8a8): 32.2 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
FCN-ResNet50	ONNX	float	Snapdragon® X2 Elite	22.125 ms	209 - 209 MB	NPU
FCN-ResNet50	ONNX	float	Snapdragon® X Elite	43.533 ms	146 - 146 MB	NPU
FCN-ResNet50	ONNX	float	Snapdragon® 8 Gen 3 Mobile	32.689 ms	0 - 370 MB	NPU
FCN-ResNet50	ONNX	float	Snapdragon® 8 Gen 1 Mobile	83.156 ms	4 - 265 MB	NPU
FCN-ResNet50	ONNX	float	Qualcomm® QCS8550 (Proxy)	43.4 ms	3 - 7 MB	NPU
FCN-ResNet50	ONNX	float	Qualcomm® QCS8450	83.156 ms	4 - 265 MB	NPU
FCN-ResNet50	ONNX	float	Snapdragon® 8 Elite Mobile	26.745 ms	2 - 310 MB	NPU
FCN-ResNet50	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	20.957 ms	1 - 322 MB	NPU
FCN-ResNet50	ONNX	float	Qualcomm® QCS9075	76.114 ms	3 - 48 MB	NPU
FCN-ResNet50	ONNX	float	Qualcomm® QCS8750	26.745 ms	2 - 310 MB	NPU
FCN-ResNet50	ONNX	float	Qualcomm® QCS7181	43.533 ms	146 - 146 MB	NPU
FCN-ResNet50	ONNX	w8a8	Snapdragon® X2 Elite	7.284 ms	212 - 212 MB	NPU
FCN-ResNet50	ONNX	w8a8	Snapdragon® X Elite	13.768 ms	153 - 153 MB	NPU
FCN-ResNet50	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	10.036 ms	1 - 267 MB	NPU
FCN-ResNet50	ONNX	w8a8	Snapdragon® 8 Gen 1 Mobile	20.949 ms	1 - 262 MB	NPU
FCN-ResNet50	ONNX	w8a8	Qualcomm® QCS6490	92.479 ms	1 - 47 MB	NPU
FCN-ResNet50	ONNX	w8a8	Qualcomm® QCS8550 (Proxy)	13.656 ms	0 - 54 MB	NPU
FCN-ResNet50	ONNX	w8a8	Qualcomm® QCS8450	20.949 ms	1 - 262 MB	NPU
FCN-ResNet50	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	7.437 ms	1 - 248 MB	NPU
FCN-ResNet50	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	25.994 ms	1 - 279 MB	NPU
FCN-ResNet50	ONNX	w8a8	Qualcomm® QCM6690	412.098 ms	0 - 380 MB	NPU
FCN-ResNet50	ONNX	w8a8	Qualcomm® QCS9075	15.091 ms	1 - 47 MB	NPU
FCN-ResNet50	ONNX	w8a8	Snapdragon® 8 Elite Mobile	8.757 ms	0 - 204 MB	NPU
FCN-ResNet50	ONNX	w8a8	Qualcomm® QCS7790	25.994 ms	1 - 279 MB	NPU
FCN-ResNet50	ONNX	w8a8	Qualcomm® QCS8750	8.757 ms	0 - 204 MB	NPU
FCN-ResNet50	ONNX	w8a8	Qualcomm® QCS7181	13.768 ms	153 - 153 MB	NPU
FCN-ResNet50	QNN_DLC	float	Snapdragon® X2 Elite	22.755 ms	3 - 3 MB	NPU
FCN-ResNet50	QNN_DLC	float	Snapdragon® X Elite	44.686 ms	3 - 3 MB	NPU
FCN-ResNet50	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	33.313 ms	1 - 377 MB	NPU
FCN-ResNet50	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	83.727 ms	4 - 274 MB	NPU
FCN-ResNet50	QNN_DLC	float	Qualcomm® QCS8275	273.613 ms	1 - 309 MB	NPU
FCN-ResNet50	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	44.404 ms	3 - 6 MB	NPU
FCN-ResNet50	QNN_DLC	float	Qualcomm® QCS8450	83.727 ms	4 - 274 MB	NPU
FCN-ResNet50	QNN_DLC	float	Snapdragon® 8 Elite Mobile	26.58 ms	3 - 324 MB	NPU
FCN-ResNet50	QNN_DLC	float	Qualcomm® SA8295P	77.798 ms	1 - 219 MB	NPU
FCN-ResNet50	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	21.314 ms	3 - 337 MB	NPU
FCN-ResNet50	QNN_DLC	float	Qualcomm® SA7255P	273.613 ms	1 - 309 MB	NPU
FCN-ResNet50	QNN_DLC	float	Qualcomm® QCS9075	78.48 ms	3 - 8 MB	NPU
FCN-ResNet50	QNN_DLC	float	Qualcomm® QCS8750	26.58 ms	3 - 324 MB	NPU
FCN-ResNet50	QNN_DLC	float	Qualcomm® QCS7181	44.686 ms	3 - 3 MB	NPU
FCN-ResNet50	QNN_DLC	w8a8	Snapdragon® X2 Elite	7.791 ms	1 - 1 MB	NPU
FCN-ResNet50	QNN_DLC	w8a8	Snapdragon® X Elite	15.214 ms	1 - 1 MB	NPU
FCN-ResNet50	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	11.283 ms	1 - 265 MB	NPU
FCN-ResNet50	QNN_DLC	w8a8	Snapdragon® 8 Gen 1 Mobile	24.389 ms	1 - 264 MB	NPU
FCN-ResNet50	QNN_DLC	w8a8	Qualcomm® QCS6490	93.458 ms	1 - 3 MB	NPU
FCN-ResNet50	QNN_DLC	w8a8	Qualcomm® QCS8275	39.493 ms	1 - 207 MB	NPU
FCN-ResNet50	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	14.868 ms	1 - 153 MB	NPU
FCN-ResNet50	QNN_DLC	w8a8	Qualcomm® QCS8450	24.389 ms	1 - 264 MB	NPU
FCN-ResNet50	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	7.446 ms	1 - 248 MB	NPU
FCN-ResNet50	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	26.055 ms	1 - 280 MB	NPU
FCN-ResNet50	QNN_DLC	w8a8	Qualcomm® QCM6690	418.038 ms	1 - 385 MB	NPU
FCN-ResNet50	QNN_DLC	w8a8	Qualcomm® QCS9075	17.531 ms	3 - 5 MB	NPU
FCN-ResNet50	QNN_DLC	w8a8	Qualcomm® SA7255P	39.493 ms	1 - 207 MB	NPU
FCN-ResNet50	QNN_DLC	w8a8	Snapdragon® 8 Elite Mobile	8.942 ms	1 - 202 MB	NPU
FCN-ResNet50	QNN_DLC	w8a8	Qualcomm® SA8295P	21.902 ms	1 - 210 MB	NPU
FCN-ResNet50	QNN_DLC	w8a8	Qualcomm® QCS7790	26.055 ms	1 - 280 MB	NPU
FCN-ResNet50	QNN_DLC	w8a8	Qualcomm® QCS8750	8.942 ms	1 - 202 MB	NPU
FCN-ResNet50	QNN_DLC	w8a8	Qualcomm® QCS7181	15.214 ms	1 - 1 MB	NPU
FCN-ResNet50	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	32.916 ms	0 - 426 MB	NPU
FCN-ResNet50	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	84.005 ms	1 - 321 MB	NPU
FCN-ResNet50	TFLITE	float	Qualcomm® QCS8275	273.522 ms	1 - 354 MB	NPU
FCN-ResNet50	TFLITE	float	Qualcomm® QCS8550 (Proxy)	43.898 ms	0 - 3 MB	NPU
FCN-ResNet50	TFLITE	float	Qualcomm® SA8775P	2273.537 ms	20 - 41 MB	CPU
FCN-ResNet50	TFLITE	float	Qualcomm® SA8650P	2273.537 ms	20 - 41 MB	CPU
FCN-ResNet50	TFLITE	float	Qualcomm® SA8255P	2273.537 ms	20 - 41 MB	CPU
FCN-ResNet50	TFLITE	float	Qualcomm® QCS8450	84.005 ms	1 - 321 MB	NPU
FCN-ResNet50	TFLITE	float	Snapdragon® 8 Elite Mobile	26.944 ms	0 - 356 MB	NPU
FCN-ResNet50	TFLITE	float	Qualcomm® SA8295P	77.801 ms	0 - 267 MB	NPU
FCN-ResNet50	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	20.83 ms	0 - 373 MB	NPU
FCN-ResNet50	TFLITE	float	Qualcomm® SA7255P	273.522 ms	1 - 354 MB	NPU
FCN-ResNet50	TFLITE	float	Qualcomm® QCS9075	77.883 ms	0 - 71 MB	NPU
FCN-ResNet50	TFLITE	float	Qualcomm® QCS8750	26.944 ms	0 - 356 MB	NPU
FCN-ResNet50	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	11.003 ms	0 - 267 MB	NPU
FCN-ResNet50	TFLITE	w8a8	Snapdragon® 8 Gen 1 Mobile	23.133 ms	0 - 266 MB	NPU
FCN-ResNet50	TFLITE	w8a8	Qualcomm® QCS6490	95.208 ms	0 - 39 MB	NPU
FCN-ResNet50	TFLITE	w8a8	Qualcomm® QCS8275	38.242 ms	0 - 207 MB	NPU
FCN-ResNet50	TFLITE	w8a8	Qualcomm® QCS8550 (Proxy)	14.291 ms	0 - 3 MB	NPU
FCN-ResNet50	TFLITE	w8a8	Qualcomm® SA8775P	550.194 ms	6 - 30 MB	CPU
FCN-ResNet50	TFLITE	w8a8	Qualcomm® SA8650P	550.194 ms	6 - 30 MB	CPU
FCN-ResNet50	TFLITE	w8a8	Qualcomm® SA8255P	550.194 ms	6 - 30 MB	CPU
FCN-ResNet50	TFLITE	w8a8	Qualcomm® QCS8450	23.133 ms	0 - 266 MB	NPU
FCN-ResNet50	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	7.385 ms	0 - 246 MB	NPU
FCN-ResNet50	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	28.01 ms	0 - 297 MB	NPU
FCN-ResNet50	TFLITE	w8a8	Qualcomm® QCM6690	442.766 ms	0 - 389 MB	NPU
FCN-ResNet50	TFLITE	w8a8	Qualcomm® QCS9075	15.251 ms	0 - 35 MB	NPU
FCN-ResNet50	TFLITE	w8a8	Qualcomm® SA7255P	38.242 ms	0 - 207 MB	NPU
FCN-ResNet50	TFLITE	w8a8	Snapdragon® 8 Elite Mobile	8.615 ms	0 - 198 MB	NPU
FCN-ResNet50	TFLITE	w8a8	Qualcomm® SA8295P	21.28 ms	0 - 210 MB	NPU
FCN-ResNet50	TFLITE	w8a8	Qualcomm® QCS7790	28.01 ms	0 - 297 MB	NPU
FCN-ResNet50	TFLITE	w8a8	Qualcomm® QCS8750	8.615 ms	0 - 198 MB	NPU

License

• The license for the original implementation of FCN-ResNet50 can be found here.

References

• Fully Convolutional Networks for Semantic Segmentation
• Source Model Implementation

Community

• Join our AI Hub Slack community to collaborate, post questions and learn more about on-device AI.
• For questions or feedback please reach out to us.