1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
|
# List of Hosted Models
## Image classification (Float Models)
Model Name | Paper_Model_Files^ | Model_Size | Top-1 Accuracy | Top-5 Accuracy | TF Lite Performance^^ | Tensorflow Performance
------------------- | :---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | ---------: | -------------: | -------------: | --------------------: | ---------------------:
DenseNet | [paper](https://arxiv.org/abs/1608.06993), [tflite&pb](https://storage.googleapis.com/download.tensorflow.org/models/tflite/model_zoo/upload_20180427/densenet_2018_04_27.tgz) | 43.6 Mb | 64.2% | 85.6% | 894 ms | 1262 ms
SqueezeNet | [paper](https://arxiv.org/abs/1602.07360), [tflite&pb](https://storage.googleapis.com/download.tensorflow.org/models/tflite/model_zoo/upload_20180427/squeezenet_2018_04_27.tgz) | 5.0 Mb | 49.0% | 72.9% | 224 ms | 255 ms
NASNet mobile | [paper](https://arxiv.org/abs/1707.07012), [tflite&pb](https://storage.googleapis.com/download.tensorflow.org/models/tflite/model_zoo/upload_20180427/nasnet_mobile_2018_04_27.tgz) | 21.4 Mb | 72.2% | 90.6% | 261 ms | 389 ms
NASNet large | [paper](https://arxiv.org/abs/1707.07012), [tflite&pb](https://storage.googleapis.com/download.tensorflow.org/models/tflite/model_zoo/upload_20180427/nasnet_large_2018_04_27.tgz) | 355.3 Mb | 82.1% | 95.8% | 6697 ms | 7940 ms
ResNet_V2_50 | [paper](https://arxiv.org/abs/1603.05027), [tflite&pb](https://storage.googleapis.com/download.tensorflow.org/models/tflite/model_zoo/upload_20180427/resnet_v2_50_2018_04_27.tgz) | 102.3 Mb | 68.1% | 88.4% | 942 ms | 1008 ms
ResNet_V2_101 | [paper](https://arxiv.org/abs/1603.05027), [tflite&pb](https://storage.googleapis.com/download.tensorflow.org/models/tflite/model_zoo/upload_20180427/resnet_v2_101_2018_04_27.tgz) | 178.3 Mb | 70.4% | 89.6% | 1880 ms | 1970 ms
Inception_V3 | [paper](http://arxiv.org/abs/1512.00567), [tflite&pb](https://storage.googleapis.com/download.tensorflow.org/models/tflite/model_zoo/upload_20180427/inception_v3_2018_04_27.tgz) | 95.3 Mb | 76.9% | 93.5% | 1433 ms | 1522 ms
Inception_V4 | [paper](http://arxiv.org/abs/1602.07261), [tflite&pb](https://storage.googleapis.com/download.tensorflow.org/models/tflite/model_zoo/upload_20180427/inception_v4_2018_04_27.tgz) | 170.7 Mb | 79.6% | 94.6% | 2986 ms | 3139 ms
Inception_ResNet_V2 | [paper](https://arxiv.org/abs/1602.07261), [tflite&pb](https://storage.googleapis.com/download.tensorflow.org/models/tflite/model_zoo/upload_20180427/inception_resnet_v2_2018_04_27.tgz) | 121.0 Mb | 76.8% | 93.5% | 2731 ms | 2926 ms
Mobilenet_0.25_128 | [paper](https://arxiv.org/pdf/1704.04861.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.25_128.tgz) | 1.9 Mb | 41.5% | 66.3% | 6.2 ms | 13.0 ms
Mobilenet_0.25_160 | [paper](https://arxiv.org/pdf/1704.04861.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.25_160.tgz) | 1.9 Mb | 45.5% | 70.3% | 8.6 ms | 19.5 ms
Mobilenet_0.25_192 | [paper](https://arxiv.org/pdf/1704.04861.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.25_192.tgz) | 1.9 Mb | 47.7% | 72.3% | 12.1 ms | 27.8 ms
Mobilenet_0.25_224 | [paper](https://arxiv.org/pdf/1704.04861.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.25_224.tgz) | 1.9 Mb | 49.8% | 74.2% | 16.2 ms | 37.3 ms
Mobilenet_0.50_128 | [paper](https://arxiv.org/pdf/1704.04861.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.5_128.tgz) | 5.3 Mb | 56.3% | 79.4% | 18.1 ms | 29.9 ms
Mobilenet_0.50_160 | [paper](https://arxiv.org/pdf/1704.04861.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.5_160.tgz) | 5.3 Mb | 59.1% | 81.9% | 26.8 ms | 45.9 ms
Mobilenet_0.50_192 | [paper](https://arxiv.org/pdf/1704.04861.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.5_192.tgz) | 5.3 Mb | 61.7% | 83.6% | 35.6 ms | 65.3 ms
Mobilenet_0.50_224 | [paper](https://arxiv.org/pdf/1704.04861.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.5_224.tgz) | 5.3 Mb | 63.3% | 84.9% | 47.6 ms | 164.2 ms
Mobilenet_0.75_128 | [paper](https://arxiv.org/pdf/1704.04861.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.75_128.tgz) | 10.3 Mb | 62.1% | 83.9% | 34.6 ms | 48.7 ms
Mobilenet_0.75_160 | [paper](https://arxiv.org/pdf/1704.04861.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.75_160.tgz) | 10.3 Mb | 65.3% | 86.0% | 51.3 ms | 75.2 ms
Mobilenet_0.75_192 | [paper](https://arxiv.org/pdf/1704.04861.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.75_192.tgz) | 10.3 Mb | 67.2% | 87.3% | 71.7 ms | 107.0 ms
Mobilenet_0.75_224 | [paper](https://arxiv.org/pdf/1704.04861.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.75_224.tgz) | 10.3 Mb | 68.4% | 88.2% | 95.7 ms | 143.4 ms
Mobilenet_1.0_128 | [paper](https://arxiv.org/pdf/1704.04861.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_1.0_128.tgz) | 16.9 Mb | 65.2% | 85.8% | 57.4 ms | 76.8 ms
Mobilenet_1.0_160 | [paper](https://arxiv.org/pdf/1704.04861.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_1.0_160.tgz) | 16.9 Mb | 68.0% | 87.7% | 86.0 ms | 117.7 ms
Mobilenet_1.0_192 | [paper](https://arxiv.org/pdf/1704.04861.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_1.0_192.tgz) | 16.9 Mb | 70.0% | 89.2% | 118.6 ms | 167.3 ms
Mobilenet_1.0_224 | [paper](https://arxiv.org/pdf/1704.04861.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_1.0_224.tgz) | 16.9 Mb | 70.9% | 89.9% | 160.1 ms | 224.3 ms
^ The model files include both TF Lite FlatBuffer and Tensorflow frozen Graph.
^^ The performance numbers are generated in the benchmark on Pixel-2 using
single thread large core.
## Image classification (Quantized Models)
Model Name | Paper_Model_Files | Model_Size | Top-1 Accuracy | Top-5 Accuracy | TF Lite Performance
------------------------ | :-------------------------------------------------------------------------------------------------------------------------------------------------------: | ---------: | -------------: | -------------: | ------------------:
Mobilenet_0.25_128_quant | [paper](https://arxiv.org/pdf/1712.05877.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.25_128_quant.tgz) | 0.5 Mb | 39.9% | 65.8% | 3.7 ms
Mobilenet_0.25_160_quant | [paper](https://arxiv.org/pdf/1712.05877.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.25_160_quant.tgz) | 0.5 Mb | 43.5% | 69.1% | 5.5 ms
Mobilenet_0.25_192_quant | [paper](https://arxiv.org/pdf/1712.05877.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.25_192_quant.tgz) | 0.5 Mb | 45.8% | 71.9% | 7.9 ms
Mobilenet_0.25_224_quant | [paper](https://arxiv.org/pdf/1712.05877.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.25_224_quant.tgz) | 0.5 Mb | 48.2% | 73.8% | 10.4 ms
Mobilenet_0.50_128_quant | [paper](https://arxiv.org/pdf/1712.05877.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.5_128_quant.tgz) | 1.4 Mb | 54.9% | 78.9% | 8.8 ms
Mobilenet_0.50_160_quant | [paper](https://arxiv.org/pdf/1712.05877.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.5_160_quant.tgz) | 1.4 Mb | 57.7% | 81.3% | 13.0 ms
Mobilenet_0.50_192_quant | [paper](https://arxiv.org/pdf/1712.05877.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.5_192_quant.tgz) | 1.4 Mb | 60.4% | 83.2% | 18.3 ms
Mobilenet_0.50_224_quant | [paper](https://arxiv.org/pdf/1712.05877.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.5_224_quant.tgz) | 1.4 Mb | 62.2% | 84.5% | 24.7 ms
Mobilenet_0.75_128_quant | [paper](https://arxiv.org/pdf/1712.05877.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.75_128_quant.tgz) | 2.6 Mb | 59.8% | 82.8% | 16.2 ms
Mobilenet_0.75_160_quant | [paper](https://arxiv.org/pdf/1712.05877.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.75_160_quant.tgz) | 2.6 Mb | 63.9% | 85.5% | 24.3 ms
Mobilenet_0.75_192_quant | [paper](https://arxiv.org/pdf/1712.05877.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.75_192_quant.tgz) | 2.6 Mb | 66.2% | 87.1% | 33.8 ms
Mobilenet_0.75_224_quant | [paper](https://arxiv.org/pdf/1712.05877.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_0.75_224_quant.tgz) | 2.6 Mb | 67.9% | 88.1% | 45.4 ms
Mobilenet_1.0_128_quant | [paper](https://arxiv.org/pdf/1712.05877.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_1.0_128_quant.tgz) | 4.3 Mb | 64.0% | 85.5% | 24.9 ms
Mobilenet_1.0_160_quant | [paper](https://arxiv.org/pdf/1712.05877.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_1.0_160_quant.tgz) | 4.3 Mb | 67.3% | 87.7% | 37.4 ms
Mobilenet_1.0_192_quant | [paper](https://arxiv.org/pdf/1712.05877.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_1.0_192_quant.tgz) | 4.3 Mb | 69.0% | 88.9% | 51.9 ms
Mobilenet_1.0_224_quant | [paper](https://arxiv.org/pdf/1712.05877.pdf), [tflite&pb](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_1.0_224_quant.tgz) | 4.3 Mb | 69.7% | 89.5% | 70.2 ms
## Other models
Model | TF Lite FlatBuffer
----------------------- | :----------------:
Smart Reply 1.0 Android | [reference](https://research.googleblog.com/2017/11/on-device-conversational-modeling-with.html), [tflite](https://storage.googleapis.com/download.tensorflow.org/models/smartreply_1.0_2017_11_01.zip)
|
