Saoirse Stewart [Wed, 27 Feb 2019 11:07:57 +0000 (11:07 +0000)]
IVGCVSW-2598 Fix for constant axis issue for Tensorflow Parser
Change-Id: I8b081012529aed8e434273259c5a5ef7dc3afff7
Signed-off-by: Finn Williams <finn.williams@arm.com>
Signed-off-by: Saoirse Stewart <saoirse.stewart@arm.com>
Matthew Bentham [Wed, 27 Feb 2019 11:51:40 +0000 (11:51 +0000)]
IVGCVSW-2764 Strengthen test for thread-safety in Profiler
Replace assert with explicit conditions so that the checks
are correctly performed in release builds.
Change-Id: Id69ab3cc0aa8c61021642b7250e30c255f144a3a
Signed-off-by: Matthew Bentham <matthew.bentham@arm.com>
Ferran Balaguer [Wed, 27 Feb 2019 09:42:06 +0000 (09:42 +0000)]
IVGCVSW-2632 Fix RefMerger from QAsymm8 types with different quantization parameters
Change-Id: Ie67ce4966c5e5fef618876b027292da429de1485
Signed-off-by: Ferran Balaguer <ferran.balaguer@arm.com>
Narumol Prangnawarat [Tue, 26 Feb 2019 17:51:08 +0000 (17:51 +0000)]
IVGCVSW-2560 Add support of SSD MobileNet to TensorFlowLiteSupport.md
* Add support for Custom operator - TFLite_Detection_PostProcess
* Add Quantized SSD MobileNet to tested networks
Change-Id: Id7e729ed02dbc401638b9832c8b9b5d71a6d31b7
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Narumol Prangnawarat [Mon, 25 Feb 2019 17:26:05 +0000 (17:26 +0000)]
IVGCVSW-2560 Verify Inference test for TensorFlow Lite MobileNet SSD
* Assign output shape of MobileNet SSD to ArmNN network
* Add m_OverridenOutputShapes to TfLiteParser to set shape in GetNetworkOutputBindingInfo
* Use input quantization instead of output quantization params
* Correct data and datatype in Inference test
Change-Id: I01ac2e07ed08e8928ba0df33a4847399e1dd8394
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Nattapat Chaimanowong [Tue, 26 Feb 2019 15:53:02 +0000 (15:53 +0000)]
IVGCVSW-2708 Add Serializer and Deserializer SpaceToBatchNd
Change-Id: I14f2a660654dc87939fe413fafa9384f09594f68
Signed-off-by: Nattapat Chaimanowong <nattapat.chaimanowong@arm.com>
keidav01 [Tue, 26 Feb 2019 11:57:39 +0000 (11:57 +0000)]
IVGCVSW-2429 Add Detection PostProcess Parser to TensorFlow Lite Parser
* Added additional custom option use_regular_nms to JSON string
* Renamed use_regular_non_max_suppression to use_regular_nms in parser
Change-Id: I5431eabade6337f6f8c3372670862e32e9cda63c
Signed-off-by: keidav01 <keith.davis@arm.com>
Conor Kennedy [Tue, 26 Feb 2019 08:29:54 +0000 (08:29 +0000)]
IVGCVSW-2683 Add Serializer & Deserializer for Constant
Change-Id: Iad7d89dfa963d9015cbe044f67aecc8bf6634b10
Signed-off-by: Conor Kennedy <conor.kennedy@arm.com>
Éanna Ó Catháin [Tue, 26 Feb 2019 11:26:24 +0000 (11:26 +0000)]
IVGCVSW-2629 Adding support for uint8 in ExecuteNetwork to fix issues with output tensor being all zero
Change-Id: I4da1ffd684672fff0853fb053c3340d06a3bc165
Signed-off-by: Éanna Ó Catháin <eanna.ocathain@arm.com>
Nattapat Chaimanowong [Tue, 26 Feb 2019 12:26:08 +0000 (12:26 +0000)]
IVGCVSW-2756 Fix bug in ClSpaceToBatchNdWorkload
*Padding was not in format expected by Cl
Change-Id: Id0de3fd2b306e31d555e2d67bc4406e5302dcd3b
Signed-off-by: Nattapat Chaimanowong <nattapat.chaimanowong@arm.com>
Derek Lamberti [Tue, 26 Feb 2019 10:49:52 +0000 (10:49 +0000)]
IVGCVSW-2525 *Update TfLite supported operator list
Change-Id: Ia349d71ec966b37feaffb971c7a77bf8cd44501c
Signed-off-by: Derek Lamberti <derek.lamberti@arm.com>
Éanna Ó Catháin [Mon, 25 Feb 2019 16:26:29 +0000 (16:26 +0000)]
IVGCVSW-2751 When layers are Deserialized the layer name is lost
* Fixed defect
* Added unit tests to check names
Change-Id: I67211110b90aabff6e00bccdadd1145423b07523
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Signed-off-by: Éanna Ó Catháin <eanna.ocathain@arm.com>
Matthew Bentham [Mon, 25 Feb 2019 13:58:24 +0000 (13:58 +0000)]
Rename ArmNN schema file to ArmnnSchema.fbs
This makes the header file change to ArmnnSchema_generated.h,
which is then unambiguous with respect to other generated schema files.
Change-Id: Id7785ed74cced4dcd4d6bcbab81e53c6aeba973c
Signed-off-by: Matthew Bentham <matthew.bentham@arm.com>
Aron Virginas-Tar [Mon, 25 Feb 2019 14:12:46 +0000 (14:12 +0000)]
IVGCVSW-2762 Fix missing layer name and output slot index in DebugLayer output
Change-Id: Idda9c8ba244c74d931105ced90c38a12d48df48a
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Bruno Goncalves [Wed, 13 Feb 2019 00:59:22 +0000 (22:59 -0200)]
Add strided-slice parser to tf-lite
Change-Id: I1821d7e8123c76823562dd2e8822c5293fcb18c3
Signed-off-by: Bruno Goncalves <bruno.slackware@gmail.com>
Saoirse Stewart [Mon, 25 Feb 2019 09:22:58 +0000 (09:22 +0000)]
IVGCVSW-2757 Add check for wrong number of components supplied to const tensor
Change-Id: Ia9bc6c73ce246712c41496a1cfe0bb6a1d2eb8e9
Signed-off-by: Saoirse Stewart <saoirse.stewart@arm.com>
Bruno Goncalves [Wed, 13 Feb 2019 00:58:18 +0000 (22:58 -0200)]
Add minimum parser to tf-lite
Change-Id: I617c2715e0fb8d01b96fa9d3d5d9a16a33477b66
Signed-off-by: Bruno Goncalves <bruno.slackware@gmail.com>
Bruno Goncalves [Wed, 13 Feb 2019 00:57:13 +0000 (22:57 -0200)]
Add maximum parser to tf-lite
Change-Id: Idaf6dd3f4d96dad01e1dc1a0d3fd6c146780626d
Signed-off-by: Bruno Goncalves <bruno.slackware@gmail.com>
Francis Murtagh [Fri, 22 Feb 2019 16:35:13 +0000 (16:35 +0000)]
IVGCVSW-2721 Quantize and Dequantize aren't quite right
* Add check for infinity and negative infinity in quantize()
* Add assert for NaN value in quantize and dequantize()
* Add unit tests for infinity and negative infinity
Change-Id: Ie60e1e15b289ccbf99df4a3281f067b82cc9f9bf
Signed-off-by: Francis Murtagh <francis.murtagh@arm.com>
Matthew Bentham [Fri, 22 Feb 2019 17:30:32 +0000 (17:30 +0000)]
IVGCVSW-2735 Fail build if TfLite parser or armnnSerializer can't be built
Change-Id: I39a2fee99218a0a31d8ce6c46d7cc1c0ca5b75d2
Signed-off-by: Matthew Bentham <matthew.bentham@arm.com>
Bruno Goncalves [Fri, 8 Feb 2019 21:02:48 +0000 (19:02 -0200)]
Add space-to-batch-nd parser to tf-lite
Change-Id: I3bf86d44f811380559ec35eed0bc43b3bd97da80
Signed-off-by: Bruno Goncalves <bruno.slackware@gmail.com>
Bruno Goncalves [Fri, 8 Feb 2019 20:52:21 +0000 (18:52 -0200)]
Add batch-to-space-nd parser to tf-lite
Change-Id: I1e210f88f19f38ab719008e119c8fed153a3dd95
Signed-off-by: Bruno Goncalves <bruno.slackware@gmail.com>
Bruno Goncalves [Thu, 7 Feb 2019 20:40:11 +0000 (18:40 -0200)]
Add resize-bilinear parser to tf-lite
Change-Id: Id35db981b38348e5a941cfbb4cbdfe8cd617a254
Signed-off-by: Bruno Goncalves <bruno.slackware@gmail.com>
Bruno Goncalves [Thu, 7 Feb 2019 20:37:39 +0000 (18:37 -0200)]
Add sub parser to tf-lite
Change-Id: I0e63d81d95528e6c278dc5fa8b87ccc1f8798130
Signed-off-by: Bruno Goncalves <bruno.slackware@gmail.com>
Aron Virginas-Tar [Fri, 22 Feb 2019 10:53:49 +0000 (10:53 +0000)]
IVGCVSW-2589 Fix for checking out CL release branch
Change-Id: I3200cb7a88158ec3aabc6feaca452406d5408129
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Saoirse Stewart [Fri, 22 Feb 2019 09:56:18 +0000 (09:56 +0000)]
IVGCVSW-2739 Updating the Readme.md and ContributorGuide.md files to add references and links to the new MLPlatform.org website (the home for the machine learning platform
Change-Id: I14819d71157ea32bb911f0a20ece3e5d2d59f64b
Signed-off-by: Saoirse Stewart <saoirse.stewart@arm.com>
Aron Virginas-Tar [Thu, 21 Feb 2019 15:45:03 +0000 (15:45 +0000)]
IVGCVSW-2749 Throw exception in TensorShape when requested index >= number of dimensions
Change-Id: I3589b1e901b0f81f6bb17848046a22829f91bb9e
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Derek Lamberti [Thu, 21 Feb 2019 16:29:43 +0000 (16:29 +0000)]
IVGSVSW-2736 Fix dangling reference to temporary storage
Change-Id: Ie6c553798eac732f37148d81970366e5a4ede1be
Signed-off-by: Derek Lamberti <derek.lamberti@arm.com>
Nattapat Chaimanowong [Thu, 21 Feb 2019 17:05:39 +0000 (17:05 +0000)]
IVGCVSW-2588 Update README files for 19.02
*Also update related Support.md files
Change-Id: If832980fdebb136ab02333d99512fd39b9093b2b
Signed-off-by: Nattapat Chaimanowong <nattapat.chaimanowong@arm.com>
Finn Williams [Mon, 18 Feb 2019 10:07:34 +0000 (10:07 +0000)]
IVGCVSW-2424 Add LOGISTIC Operator Support (CL)
* Added Logistic tests in ClLayerTests
* Removed check for QuantisedAsymm8 in CLActivationWorkload
Change-Id: I6950e18109504d8426b727cb6b7d19898c899024
Signed-off-by: Finn Williams <finn.williams@arm.com>
Aron Virginas-Tar [Thu, 21 Feb 2019 14:57:08 +0000 (14:57 +0000)]
IVGCVSW-2748 Fix bug causing std::bad_alloc in MakeTensor()
Change-Id: Iff6533396e594e2716e432223780d662d99d06ec
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Finn Williams [Mon, 18 Feb 2019 13:07:42 +0000 (13:07 +0000)]
IVGCVSW-2589 Update CL pin to the 19.02 release
Change-Id: Ic3a0ab87ac1df94bd69a2b966b27f95a7ef47dc9
Signed-off-by: Finn Williams <finn.williams@arm.com>
Jim Flynn [Wed, 20 Feb 2019 14:25:11 +0000 (14:25 +0000)]
IVGCVSW-2635 Update TensorFlowSupport.md
* Also fixed a format issue in TfParser.hpp
Change-Id: I5b0ae1d5909f03e166a40d3aca8859024cd5e5b6
Signed-off-by: Jim Flynn <jim.flynn@arm.com>
telsoa01 [Wed, 20 Feb 2019 11:44:48 +0000 (11:44 +0000)]
COMPMID-2007 Changing NDK version to 17b in the Build ArmNN with the Android NDK guide. This is required by the Arm Compute Library for some of the v8 NEON intrinsics
Change-Id: Icd4bfaf502c4ada55618b711603d5ff509674b55
keidav01 [Thu, 21 Feb 2019 10:07:37 +0000 (10:07 +0000)]
IVGCVSW-2429 Add Detection PostProcess Parser to TensorFlow Lite Parser
* Added parser function in TFLiteParser
* Removed custom options gating
* Added unit test
* Removed template usage in VerifyTensorInfo for DeserializeParser
Change-Id: If198654ed70060855a05f8aaed010293405bd103
Signed-off-by: keidav01 <keith.davis@arm.com>
Aron Virginas-Tar [Thu, 21 Feb 2019 09:41:28 +0000 (09:41 +0000)]
IVGCVSW-2643 Add Activation to SerializerSupport.md and DeserializerSupport.md
Change-Id: Id4a75051d8a93744346b91febcad0e62e7063495
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Sadik Armagan [Thu, 21 Feb 2019 09:01:41 +0000 (09:01 +0000)]
IVGCVSW-2639 Add Serializer & Deserializer for Fully Connected
* Added FullyConnectedLayer to Serializer Schema Schema.fbs
* Added FullyConnected serialization and deserialization support
* Added FullyConnected serialization and deserialization unit tests
Change-Id: I8ef14f9728158f849fa4d1a8d05a1a4170cd5b41
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Ferran Balaguer [Mon, 18 Feb 2019 12:47:35 +0000 (12:47 +0000)]
IVGCVSW-2613 Support static quantization of BatchToSpace
Change-Id: I44b12c5c246b7aacc789420dbe55a16efaab6f98
ruoyan01 [Wed, 20 Feb 2019 13:53:22 +0000 (13:53 +0000)]
Fix typos in IVGCVSW-2645 Add Serializer & Deserializer for Pooling2d
Change-Id: I63fdf3e7fb6ba42e879942ee9664d7c84678d4e8
Signed-off-by: ruoyan01 <ruomei.yan@arm.com>
Jim Flynn [Fri, 15 Feb 2019 14:45:04 +0000 (14:45 +0000)]
IVGCVSW-2610 Add Quantization Tool Executable
* Also added TensorInfos to the output slots in the SerializerTests
to create valid ArmNN networks for serialization
Change-Id: I092b1ac889dd5e05cb1c10c9dfb573acaf1970d9
Signed-off-by: Jim Flynn <jim.flynn@arm.com>
Nattapat Chaimanowong [Wed, 20 Feb 2019 17:31:34 +0000 (17:31 +0000)]
IVGCVSW-2702 Add Serializer and Deserializer for Permute
Change-Id: I56322cb935de42cfa25bd1fbb0c09d00c7a5dd2b
Signed-off-by: Nattapat Chaimanowong <nattapat.chaimanowong@arm.com>
Mike Kelly [Wed, 20 Feb 2019 16:53:11 +0000 (16:53 +0000)]
IVGCVSW-2643 Add Serializer & Deserializer for Activation
* Added ActivationLayer to Schema.fbs
* Added Activation serialization and deserialization support
* Added serialization and deserialization unit tests
Change-Id: Ib5df45f123674988b994ffe3f111d3fb57864912
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Derek Lamberti [Wed, 20 Feb 2019 13:57:42 +0000 (13:57 +0000)]
IVGSVSW-2736 Rename DeserializeParser => Deserializer & fix namespaces
Change-Id: I4166c0bbb5ba7f8b8884e71134c21f43d1fc27b0
Signed-off-by: Derek Lamberti <derek.lamberti@arm.com>
Aron Virginas-Tar [Wed, 20 Feb 2019 13:38:47 +0000 (13:38 +0000)]
IVGCVSW-2636 Update serializer and deserializer README with new layers
Change-Id: Icba4e6e65027c4ac2dd08bd5a2662c4289ade8f3
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Aron Virginas-Tar [Tue, 19 Feb 2019 16:31:08 +0000 (16:31 +0000)]
IVGCVSW-2685 Serialize / de-serialize the DepthwiseConvolution2d layer
Change-Id: I37e360c824b30cb14cbef86f6ff7636bc9382109
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Saoirse Stewart [Wed, 20 Feb 2019 09:54:19 +0000 (09:54 +0000)]
IVGCVSW-2583 Update verison number for 19.02
Change-Id: I9c28b8986d10b092ddc8e9b609006b27858bf6be
Signed-off-by: Saoirse Stewart <saoirse.stewart@arm.com>
Narumol Prangnawarat [Fri, 15 Feb 2019 17:34:51 +0000 (17:34 +0000)]
IVGCVSW-2436 Modify MobileNet SSD inference test
* change MobileNet SSD input to uint8
* get quantization scale and offset from the model
* change data layout to NHWC as TensorFlow lite layout
* update expected output as result from TfLite with quantized data
Change-Id: I07104d56286893935779169356234de53f1c9492
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Nattapat Chaimanowong [Fri, 15 Feb 2019 16:46:24 +0000 (16:46 +0000)]
IVGCVSW-2730 Fix bug in TfParser setting incorrect output shape for Addition Layer
*Output shape was incorrectly set for case when the 2 inputs have same
number of dimensions and each dimension has different (but broadcast
compatible) size
Change-Id: Ia617442b2b7f962c33792b94035236d26de70558
Signed-off-by: Nattapat Chaimanowong <nattapat.chaimanowong@arm.com>
Saoirse Stewart [Mon, 18 Feb 2019 17:19:16 +0000 (17:19 +0000)]
IVGCVSW-2725 Fix issue with ExecuteNetwork and Guid's
Change-Id: I499b2ac3b51390fd725ea8afa64efc7e007be199
Signed-off-by: Saoirse Stewart <saoirse.stewart@arm.com>
Mike Kelly [Tue, 19 Feb 2019 17:22:07 +0000 (17:22 +0000)]
IVGCVSW-2646 Add Serializer & Deserializer for Conv2D
* Added Convolution2dLayer to Schema.fbs
* Added ConstTensorData serialization and deserialization helper functions
* Added Convolution2d serialization and deserialization support
* Added serialization and deserialization unit tests
Change-Id: Id376c08410ae01511972a2b0abdce9cfab907462
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Derek Lamberti [Mon, 18 Feb 2019 16:36:57 +0000 (16:36 +0000)]
IVGCVSW-2736 Deserialize using istream instead of filename
Change-Id: I5656b23d9783e7f953e677001d16e41eedeb42b2
Signed-off-by: Derek Lamberti <derek.lamberti@arm.com>
Saoirse Stewart [Tue, 19 Feb 2019 15:54:14 +0000 (15:54 +0000)]
IVGCVSW-2642 Add Reshape to Serializer and Deserializer
Change-Id: Iccded3c6e3c0713c44f43231981440420591f94e
Signed-off-by: Saoirse Stewart <saoirse.stewart@arm.com>
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Nattapat Chaimanowong [Thu, 14 Feb 2019 15:28:02 +0000 (15:28 +0000)]
IVGCVSW-2662 Implement conversion tool to serialize ArmNN networks
Change-Id: I86fb93b8d739ab043e019643184a091e700fd7c4
Signed-off-by: Nattapat Chaimanowong <nattapat.chaimanowong@arm.com>
Saoirse Stewart [Mon, 18 Feb 2019 15:24:53 +0000 (15:24 +0000)]
IVGCVSW-2645 Add Serializer & Deserializer for Pooling2d
Change-Id: Iba41da3cccd539a0175f2ed0ff9a8b6a23c5fb6f
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Signed-off-by: Saoirse Stewart <saoirse.stewart@arm.com>
Francis Murtagh [Mon, 18 Feb 2019 17:14:10 +0000 (17:14 +0000)]
IVGCVSW-2619 Support static quantization of StridedSlice
Change-Id: Iff0902ea2751b079563ce82fbbec6db5157c3392
Signed-off-by: Francis Murtagh <francis.murtagh@arm.com>
Aron Virginas-Tar [Wed, 13 Feb 2019 15:41:52 +0000 (15:41 +0000)]
IVGCVSW-2644 Add Serializer & Deserializer for Softmax
Change-Id: Ifea2108e173d2b602162fe53b880a68e1c715510
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Jim Flynn [Mon, 18 Feb 2019 10:42:32 +0000 (10:42 +0000)]
IVGCVSW-2626 Add Quantization of ResizeBilinear Layer
Change-Id: I0daffd23f32cc094d8309aed822cc46bb2aaa46f
Signed-off-by: Jim Flynn <jim.flynn@arm.com>
Francis Murtagh [Fri, 15 Feb 2019 16:56:27 +0000 (16:56 +0000)]
IVGCVSW-2618 Support static quantization of Splitter
Change-Id: I8e44866336dcd5a9694309cf9bb954a4991e61fd
Signed-off-by: Francis Murtagh <francis.murtagh@arm.com>
Nina Drozd [Thu, 14 Feb 2019 13:41:49 +0000 (13:41 +0000)]
IVGCVSW-2616: static quantization of reshape
Change-Id: Iaefa145b8a218b3708c8a9208fdabd4984376ae9
Signed-off-by: Nina Drozd <nina.drozd@arm.com>
Jim Flynn [Thu, 14 Feb 2019 09:56:36 +0000 (09:56 +0000)]
IVGCVSW-2621 Add static quantization of Merger
Change-Id: I19f01698a6f9b361cd1737d76e6ec2020fab77a6
Signed-off-by: Jim Flynn <jim.flynn@arm.com>
Nattapat Chaimanowong [Wed, 13 Feb 2019 15:56:24 +0000 (15:56 +0000)]
IVGCVSW-2724 Fix bug in DeserializeParser
* Fixed error in which DeserializeParser::LoadGraphFromFile returns
a pointer to a local variable in its scope
Change-Id: Ic928b8509c0ae76570f0527b17e050b46327fe0c
Signed-off-by: Nattapat Chaimanowong <nattapat.chaimanowong@arm.com>
Saoirse Stewart [Thu, 14 Feb 2019 15:46:10 +0000 (15:46 +0000)]
IVGCVSW-2642 Fix Guid issue in Serialize Parser
Change-Id: Id8ba083b6a5dee64ed197d71647ad3b451e4f354
Signed-off-by: Saoirse Stewart <saoirse.stewart@arm.com>
Saoirse Stewart [Thu, 14 Feb 2019 14:57:37 +0000 (14:57 +0000)]
IVGCVSW-2642 Fix layerBindingId issue
Change-Id: I24fef25ab12df07b98d78de98563ee4a2b9977ba
Signed-off-by: Saoirse Stewart <saoirse.stewart@arm.com>
Matteo Martincigh [Mon, 11 Feb 2019 13:24:38 +0000 (13:24 +0000)]
IVGCVSW-2620 Support static quantization of Constant
* Added VisitConstantLayer to QuantizerVisitor
* Added unit tests and refactored QuantizerTest.cpp
* Code cleanup
Change-Id: I118fd2be085fc98879c5cfaa09698a7c98ba13f0
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
Matteo Martincigh [Wed, 13 Feb 2019 15:12:39 +0000 (15:12 +0000)]
IVGCVSW-2726 Code cleanup
* Removed extra ';'
* Renamed file to fix typo
Change-Id: I81a25faf01bf0a285e45fc81febf36c981d25f2f
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
Matthew Bentham [Thu, 14 Feb 2019 12:49:31 +0000 (12:49 +0000)]
IVGCVSW-2658 Integrate Compute Library 6eb90d7
Change-Id: I4a7e8b2caf41bc7edcfea44b883406f3c1fa4ba0
Signed-off-by: Matthew Bentham <matthew.bentham@arm.com>
FrancisMurtagh [Wed, 13 Feb 2019 16:43:14 +0000 (16:43 +0000)]
IVGCVSW-2615 Support static quantization of Pooling2d
Change-Id: Ica9c48adad741840b201047bb65541a19aac17f7
Signed-off-by: FrancisMurtagh <francis.murtagh@arm.com>
Nattapat Chaimanowong [Tue, 12 Feb 2019 14:38:31 +0000 (14:38 +0000)]
IVGCVSW-2714 Fix armv7 build
Change-Id: I3e2286071636cb8957dca209e578204330e879e4
Signed-off-by: Nattapat Chaimanowong <nattapat.chaimanowong@arm.com>
Nattapat Chaimanowong [Wed, 13 Feb 2019 15:52:41 +0000 (15:52 +0000)]
IVGCVSW-2662 Create ISerializer class and refactor Serializer
Change-Id: I8a500182193cd218d88c2bb406bf22f496fe43d1
Signed-off-by: Nattapat Chaimanowong <nattapat.chaimanowong@arm.com>
Jim Flynn [Wed, 13 Feb 2019 12:06:46 +0000 (12:06 +0000)]
IVGCVSW-2617 Add static quantization of SpaceToBatch
* Some refactor of the existing QuantizePermute test for reuse
Change-Id: Ifaf1afc476bad348f260fa5340c0153e7ade2703
Signed-off-by: Jim Flynn <jim.flynn@arm.com>
Sadik Armagan [Tue, 12 Feb 2019 14:31:45 +0000 (14:31 +0000)]
IVGCVSW-2640 Add Serializer & Deserializer for Mul
* Updated Serializer schema for Multiplication support
* Added support for Multiplication to Serializer and Deserializer
Change-Id: I10ad8ad4b37876a963ccdcf7074cb66f40531bde
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
FrancisMurtagh [Tue, 12 Feb 2019 16:52:24 +0000 (16:52 +0000)]
IVGCVSW-2614 Support static quantization of Permute
Change-Id: I4f208d18f7ac6bb04423cc26ecf61adf0ccc5d26
Signed-off-by: FrancisMurtagh <francis.murtagh@arm.com>
Derek Lamberti [Fri, 8 Feb 2019 17:54:20 +0000 (17:54 +0000)]
IVGCVSW-2607 Refactor range tracking into own class
Change-Id: I1b409e5dac7922859e04a554893b982afc5ad1e7
Signed-off-by: Derek Lamberti <derek.lamberti@arm.com>
Aron Virginas-Tar [Tue, 12 Feb 2019 11:27:53 +0000 (11:27 +0000)]
IVGCVSW-2663 Enable ExecuteNetwork to load ArmNN files
Change-Id: I1a61a1da2258bd07b39da6063d22a5bd22c1884d
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
FrancisMurtagh [Tue, 12 Feb 2019 10:11:43 +0000 (10:11 +0000)]
IVGCVSW-2623 Support static quantization of DepthwiseConv2d
Change-Id: Iab0d5aed243aca921661e4d39770fe02b1330442
Signed-off-by: FrancisMurtagh <francis.murtagh@arm.com>
Mike Kelly [Mon, 11 Feb 2019 17:01:27 +0000 (17:01 +0000)]
IVGCVSW-2531 Serialize a simple ArmNN Network
Change-Id: I68cf5072aca6e3a8b3b8c57e19b6d417cd5813fc
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
ruoyan01 [Mon, 11 Feb 2019 13:22:32 +0000 (13:22 +0000)]
IVGCVSW-2625 Support static quantization of softmax
Change-Id: I216344ee10bbffadb648d4aef1d9a0d9dbb4a341
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Signed-off-by: ruoyan01 <ruomei.yan@arm.com>
Aron Virginas-Tar [Mon, 11 Feb 2019 12:21:27 +0000 (12:21 +0000)]
IVGCVSW-2676 Make biases optional in ILayerVisitor for Convolution2D, DepthwiseConvolution2D and FullyConnected
Change-Id: I3048504ff699fdb266488e7c07b7262e5843d4b0
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Jim Flynn [Mon, 11 Feb 2019 10:30:30 +0000 (10:30 +0000)]
IVGCVSW-2622 Add static quantization of 2DConvolution
Change-Id: If7985a54eba97f7c61413e0804879e4afbf65c4d
Signed-off-by: Jim Flynn <jim.flynn@arm.com>
FrancisMurtagh [Mon, 11 Feb 2019 11:06:33 +0000 (11:06 +0000)]
IVGCVSW-2624 Support static quantization of FullyConnected
Change-Id: Ib18085e7e4b2e9d55e29b14122410732e3bdd088
Signed-off-by: FrancisMurtagh <francis.murtagh@arm.com>
Kevin May [Fri, 8 Feb 2019 16:31:42 +0000 (16:31 +0000)]
IVGCVSW-2581 Create Deserializer
* Add deserialize parser for input, output and add layers
* Add Unit Tests for simple network
Change-Id: Ia0e2a234896bbe401ed0da5f18c065cb5df51bfb
Signed-off-by: Kevin May <kevin.may@arm.com>
Signed-off-by: Saoirse Stewart <saoirse.stewart@arm.com>
Ferran Balaguer [Fri, 8 Feb 2019 17:09:55 +0000 (17:09 +0000)]
IVGCVSW-2529 DeepSpeech v1 test
Change-Id: Ieb99ac1aa347cee4b28b831753855c4614220648
Matteo Martincigh [Thu, 7 Feb 2019 17:52:41 +0000 (17:52 +0000)]
IVGCVSW-2672 Code cleanup after changes
* Minor code cleanup and refactoring
Change-Id: I9c6390c15944686134ddf4b47839762f2bb13922
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
Matteo Martincigh [Thu, 7 Feb 2019 17:51:09 +0000 (17:51 +0000)]
IVGCVSW-2607 Implement Input range override mechanism
* Added the OverrideInputRange method to the Quantizer API
* Created OverrideInputRangeVisitor to implement the override mechanism
* Moved the quantizer utility functions to the new NetworkQuantizerUtils files
* Moved the map of quantization ranges out of the StaticRangeVisitor
and into the NetworkQuantizer
* Added unit tests
* Code refactoring and cleanup
Change-Id: I9c1d006c1b6a35fbc04584a832fbe489f8f9276d
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
Narumol Prangnawarat [Mon, 4 Feb 2019 19:05:27 +0000 (19:05 +0000)]
IVGCVSW-2557 Add layer tests for Ref Detection PostProcess
Change-Id: Ia2e512c0ed035bc6ee46cd67df5e619da9770900
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Matthew Bentham [Thu, 7 Feb 2019 11:35:09 +0000 (11:35 +0000)]
Github #123 Catch exceptions by reference
Change-Id: Ie664cb7127434ad9c7a03c4a31f3c71fdebd49b9
Signed-off-by: Matthew Bentham <Matthew.Bentham@arm.com>
Narumol Prangnawarat [Mon, 4 Feb 2019 11:46:26 +0000 (11:46 +0000)]
IVGCVSW-2559 End to end tests for Detection PostProcess
* end to end tests for Detection PostProcess float and uint8
* add anchors to AddDetectionPostProcessLayer
* add anchors to VisitDetectionPostProcessLayer
* refactor code
Change-Id: I3c5a9a4a60b74c2246b4a27692bbf3c235163f90
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Bruno Goncalves [Fri, 28 Dec 2018 12:08:26 +0000 (10:08 -0200)]
Added TfLiteParser test for MnasNet
Change-Id: Ie31eee48cc14ada37526130998da7a482d56b1ea
Jim Flynn [Thu, 7 Feb 2019 17:42:32 +0000 (17:42 +0000)]
IVGCVSW-2553 Change BOOST_TEST to BOOST_CHECK_MESSAGE in CheckConstTensorPtrs
Change-Id: I13614fe3d29eb1040b5927dd69992aab6c136bb6
Signed-off-by: Jim Flynn <jim.flynn@arm.com>
Nina Drozd [Thu, 7 Feb 2019 11:09:27 +0000 (11:09 +0000)]
IVGCVSW-2608: support static quantization of Activation
Change-Id: Ia9afd15d002d4454ec72f219c5cf214704f6ae31
Signed-off-by: Nina Drozd <nina.drozd@arm.com>
jimfly01 [Tue, 5 Feb 2019 09:20:11 +0000 (09:20 +0000)]
IVGCVSW-2553 Add Unit Test for ConstTensor layer visitors
Change-Id: I20a67135e4af7aa1f28f5000d73122f4e5e3acd5
Signed-off-by: Jim Flynn <jim.flynn@arm.com>
Derek Lamberti [Thu, 7 Feb 2019 11:14:11 +0000 (11:14 +0000)]
IVGCVSW-2609 Quantize BatchNormalizationLayer
Change-Id: I7b847112a0322ffc8b88a0708d8439bfb97cfe2c
Signed-off-by: Derek Lamberti <derek.lamberti@arm.com>
Saoirse Stewart [Tue, 5 Feb 2019 17:27:06 +0000 (17:27 +0000)]
IVGCVSW-2581 Update Schema.fbs to refer to vector of tables
Change-Id: Ib52e4b5db540093ae9290697ace246e26f99dadb
Bruno Goncalves [Fri, 28 Dec 2018 12:09:53 +0000 (10:09 -0200)]
Added TfLiteParser test for InceptionV4Quantized
Change-Id: Iad57395b0231a0960e7fc27c1d487bdee8a1f464
Bruno Goncalves [Thu, 27 Dec 2018 18:15:38 +0000 (16:15 -0200)]
Added TfLiteParser test for ResNetV2
Change-Id: If45696594503cd093c7d5f3b5001b20e9bb2e573
Derek Lamberti [Wed, 6 Feb 2019 16:20:46 +0000 (16:20 +0000)]
IVGCVSW-2654 Test quantization of addition, input and output
Change-Id: I6c61df213278e6ba65807932548bb2eaa159deba
Signed-off-by: Derek Lamberti <derek.lamberti@arm.com>
Nattapat Chaimanowong [Thu, 31 Jan 2019 15:36:39 +0000 (15:36 +0000)]
IVGCVSW-2580 Integrate serialization schema code generation to build process
Change-Id: I092df948346281878be7e0a1d31b0748d99a99b2
Signed-off-by: Nattapat Chaimanowong <nattapat.chaimanowong@arm.com>
Derek Lamberti [Tue, 5 Feb 2019 16:00:08 +0000 (16:00 +0000)]
IVGCVSW-2606 Produce quantized InputNetwork from simple FP32 InputNetwork
Change-Id: I2140a7af5961ddf8267fbb127202de3900ea79e3
Signed-off-by: Derek Lamberti <derek.lamberti@arm.com>
FrancisMurtagh [Mon, 4 Feb 2019 15:41:17 +0000 (15:41 +0000)]
IVGCVSW-2551 Add Descriptor and name only unit tests for Visitor
* Add concrete subclasses for layers taking descriptor as argument
* Add unit test cases for all layers
Change-Id: Ie0135b34a5a41e5ce654b7c3186eccea392ab919
Signed-off-by: FrancisMurtagh <francis.murtagh@arm.com>
Narumol Prangnawarat [Thu, 31 Jan 2019 15:31:54 +0000 (15:31 +0000)]
IVGCVSW-2557 Ref Workload Implementation for Detection PostProcess
* implementation of DetectionPostProcessQueueDescriptor validate
* add Uint8ToFloat32Workload
* add implementation of Detection PostProcess functionalities
* add ref workload implemenentation for float and uint8
* add layer support for Detection PostProcess in ref
* unit tests
Change-Id: I650461f49edbb3c533d68ef8700377af51bc3592