ArmNN  NotReleased
src/armnnTfLiteParser/TensorFlowLiteSupport.md
Go to the documentation of this file.
1 # TensorFlow Lite operators that the Arm NN SDK supports
2 
3 This reference guide provides a list of TensorFlow Lite operators the Arm NN SDK currently supports.
4 
5 ## Fully supported
6 
7 The Arm NN SDK TensorFlow Lite parser currently supports the following operators:
8 
9 * ADD
10 
11 * AVERAGE_POOL_2D, Supported Fused Activation: RELU , RELU6 , TANH, NONE
12 
13 * BATCH_TO_SPACE
14 
15 * CONCATENATION, Supported Fused Activation: RELU , RELU6 , TANH, NONE
16 
17 * CONV_2D, Supported Fused Activation: RELU , RELU6 , TANH, NONE
18 
19 * DEPTHWISE_CONV_2D, Supported Fused Activation: RELU , RELU6 , TANH, NONE
20 
21 * DEQUANTIZE
22 
23 * FULLY_CONNECTED, Supported Fused Activation: RELU , RELU6 , TANH, NONE
24 
25 * LOGISTIC
26 
27 * L2_NORMALIZATION
28 
29 * MAX_POOL_2D, Supported Fused Activation: RELU , RELU6 , TANH, NONE
30 
31 * MAXIMUM
32 
33 * MEAN
34 
35 * MINIMUM
36 
37 * MUL
38 
39 * PACK
40 
41 * PAD
42 
43 * QUANTIZE
44 
45 * RELU
46 
47 * RELU6
48 
49 * RESHAPE
50 
51 * RESIZE_BILINEAR
52 
53 * RESIZE_NEAREST_NEIGHBOR
54 
55 * SLICE
56 
57 * SOFTMAX
58 
59 * SPACE_TO_BATCH
60 
61 * SPLIT
62 
63 * SQUEEZE
64 
65 * STRIDED_SLICE
66 
67 * SUB
68 
69 * TANH
70 
71 * TRANSPOSE
72 
73 * TRANSPOSE_CONV
74 
75 * UNPACK
76 
77 ## Custom Operator
78 
79 * TFLite_Detection_PostProcess
80 
81 ## Tested networks
82 
83 Arm tested these operators with the following TensorFlow Lite neural network:
84 
85 * [Quantized MobileNet](http://download.tensorflow.org/models/mobilenet_v1_2018_02_22/mobilenet_v1_1.0_224_quant.tgz)
86 
87 * [Quantized SSD MobileNet](http://download.tensorflow.org/models/object_detection/ssd_mobilenet_v1_quantized_300x300_coco14_sync_2018_07_18.tar.gz)
88 
89 * DeepSpeech v1 converted from [TensorFlow model](https://github.com/mozilla/DeepSpeech/releases/tag/v0.4.1)
90 
91 * DeepSpeaker
92 
93 More machine learning operators will be supported in future releases.