[go: up one dir, main page]
Scribd
Upload
13 views1 page

Model Performance Summary Cleaned

The document presents various models with their specifications including model size, number of parameters, mean Average Precision (mAP), datasets used, power consumption, inference frameworks, and quantization approaches. It highlights models with sizes ranging from 7 MB to 34 MB and varying power consumption, primarily using custom datasets. Several models utilize TensorFlow Lite and OpenVINO frameworks, with some employing quantization techniques for optimization.

Uploaded by

archanaonlineexam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
13 views1 page

Model Performance Summary Cleaned

The document presents various models with their specifications including model size, number of parameters, mean Average Precision (mAP), datasets used, power consumption, inference frameworks, and quantization approaches. It highlights models with sizes ranging from 7 MB to 34 MB and varying power consumption, primarily using custom datasets. Several models utilize TensorFlow Lite and OpenVINO frameworks, with some employing quantization techniques for optimization.

Uploaded by

archanaonlineexam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 1

Model Size (MB) No.

of Parameters (M) mAP (%) Dataset Used Power Consumption (W)Inference Framework Quantization Ap

Not reported Not reported mAP50=0.992, mAP50-95=0.902


Custom 2,292-image dataset
~5 W overall Not stated No

~34 MB ~8.7 M AP50 used Not reported ~5 W OpenVINO No

~17 MB (TFLite .tflite) ~3.4 M Not reported Custom robot video sequences
~5 W TensorFlow Lite; PyTorch
Yes
(on(TFLite
host) INT

~17 MB (TFLite) ~3.4 M Not reported Custom office video scenes


~3-5 W TensorFlow Lite (quantized)
Yes (INT8)

Not reported Not reported Not reported Custom dataset (not named)
~5 W TensorFlow No

~7 2.87 - MS COCO - TFLite + ARM NN Yes

~14 7.03 - MS COCO ~2 PyCoral + TFLite Yes

6.4 MB (INT8 quantized Not


model)
applicable Not explicitly reported Custom aerial emergency
4.0dataset
W TensorFlow Lite Yes, INT8 post-

Not reported Not reported Not reported Custom colored ball dataset
~5 W Likely TF/PyTorch No

0.844 Custom dataset created by the authors for indoorNCNN


drone testing None

You might also like