forked from apache/mxnet
-
Notifications
You must be signed in to change notification settings - Fork 4
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
[MXNET-50] Scala Inference APIs (apache#9678)
* Scala Inference APIs * fix unit tests for shape.length == layout.length in DataDesc * make ThreadPoolHandler of size 1 * Rename PredictBase to Predictor * change classify output from List to IndexedSeq * modify MXNetHandler to check if the task is executing on the same thread that created the handler * add argument epoch for Predictor/Classifier
- Loading branch information
Showing
12 changed files
with
946 additions
and
16 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,84 @@ | ||
| <?xml version="1.0" encoding="UTF-8"?> | ||
| <project xmlns="http://maven.apache.org/POM/4.0.0" | ||
| xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" | ||
| xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd"> | ||
| <parent> | ||
| <artifactId>mxnet-parent_2.11</artifactId> | ||
| <groupId>ml.dmlc.mxnet</groupId> | ||
| <version>1.2.0-SNAPSHOT</version> | ||
| </parent> | ||
| <modelVersion>4.0.0</modelVersion> | ||
|
|
||
| <artifactId>mxnet-infer</artifactId> | ||
| <name>MXNet Scala Package - Inference</name> | ||
|
|
||
| <profiles> | ||
| <profile> | ||
| <id>osx-x86_64-cpu</id> | ||
| <properties> | ||
| <platform>osx-x86_64-cpu</platform> | ||
| </properties> | ||
| </profile> | ||
| <profile> | ||
| <id>linux-x86_64-cpu</id> | ||
| <properties> | ||
| <platform>linux-x86_64-cpu</platform> | ||
| </properties> | ||
| </profile> | ||
| <profile> | ||
| <id>linux-x86_64-gpu</id> | ||
| <properties> | ||
| <platform>linux-x86_64-gpu</platform> | ||
| </properties> | ||
| </profile> | ||
| </profiles> | ||
|
|
||
| <build> | ||
| <plugins> | ||
| <plugin> | ||
| <groupId>org.apache.maven.plugins</groupId> | ||
| <artifactId>maven-jar-plugin</artifactId> | ||
| <configuration> | ||
| <excludes> | ||
| <exclude>META-INF/*.SF</exclude> | ||
| <exclude>META-INF/*.DSA</exclude> | ||
| <exclude>META-INF/*.RSA</exclude> | ||
| </excludes> | ||
| </configuration> | ||
| </plugin> | ||
| <plugin> | ||
| <groupId>org.apache.maven.plugins</groupId> | ||
| <artifactId>maven-compiler-plugin</artifactId> | ||
| </plugin> | ||
| <plugin> | ||
| <groupId>org.scalatest</groupId> | ||
| <artifactId>scalatest-maven-plugin</artifactId> | ||
| <configuration> | ||
| <argLine> | ||
| -Djava.library.path=${project.parent.basedir}/native/${platform}/target \ | ||
| -Dlog4j.configuration=file://${project.basedir}/src/test/resources/log4j.properties | ||
| </argLine> | ||
| </configuration> | ||
| </plugin> | ||
| <plugin> | ||
| <groupId>org.scalastyle</groupId> | ||
| <artifactId>scalastyle-maven-plugin</artifactId> | ||
| </plugin> | ||
| </plugins> | ||
| </build> | ||
| <dependencies> | ||
| <dependency> | ||
| <groupId>ml.dmlc.mxnet</groupId> | ||
| <artifactId>mxnet-core_${scala.binary.version}</artifactId> | ||
| <version>1.2.0-SNAPSHOT</version> | ||
| <scope>provided</scope> | ||
| </dependency> | ||
| <!-- https://mvnrepository.com/artifact/org.mockito/mockito-all --> | ||
| <dependency> | ||
| <groupId>org.mockito</groupId> | ||
| <artifactId>mockito-all</artifactId> | ||
| <version>1.10.19</version> | ||
| <scope>test</scope> | ||
| </dependency> | ||
| </dependencies> | ||
| </project> |
170 changes: 170 additions & 0 deletions
170
scala-package/infer/src/main/scala/ml/dmlc/mxnet/infer/Classifier.scala
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,170 @@ | ||
| /* | ||
| * Licensed to the Apache Software Foundation (ASF) under one or more | ||
| * contributor license agreements. See the NOTICE file distributed with | ||
| * this work for additional information regarding copyright ownership. | ||
| * The ASF licenses this file to You under the Apache License, Version 2.0 | ||
| * (the "License"); you may not use this file except in compliance with | ||
| * the License. You may obtain a copy of the License at | ||
| * | ||
| * http://www.apache.org/licenses/LICENSE-2.0 | ||
| * | ||
| * Unless required by applicable law or agreed to in writing, software | ||
| * distributed under the License is distributed on an "AS IS" BASIS, | ||
| * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. | ||
| * See the License for the specific language governing permissions and | ||
| * limitations under the License. | ||
| */ | ||
|
|
||
| package ml.dmlc.mxnet.infer | ||
|
|
||
| import ml.dmlc.mxnet.{Context, DataDesc, NDArray} | ||
| import java.io.File | ||
|
|
||
| import org.slf4j.LoggerFactory | ||
|
|
||
| import scala.io | ||
| import scala.collection.mutable.ListBuffer | ||
|
|
||
| trait ClassifierBase { | ||
|
|
||
| /** | ||
| * Takes an Array of Floats and returns corresponding labels, score tuples. | ||
| * @param input: IndexedSequence one-dimensional array of Floats. | ||
| * @param topK: (Optional) How many top_k(sorting will be based on the last axis) | ||
| * elements to return, if not passed returns unsorted output. | ||
| * @return IndexedSequence of (Label, Score) tuples. | ||
| */ | ||
| def classify(input: IndexedSeq[Array[Float]], | ||
| topK: Option[Int] = None): IndexedSeq[(String, Float)] | ||
|
|
||
| /** | ||
| * Takes a Sequence of NDArrays and returns Label, Score tuples. | ||
| * @param input: Indexed Sequence of NDArrays | ||
| * @param topK: (Optional) How many top_k(sorting will be based on the last axis) | ||
| * elements to return, if not passed returns unsorted output. | ||
| * @return Traversable Sequence of (Label, Score) tuple | ||
| */ | ||
| def classifyWithNDArray(input: IndexedSeq[NDArray], | ||
| topK: Option[Int] = None): IndexedSeq[IndexedSeq[(String, Float)]] | ||
| } | ||
|
|
||
| /** | ||
| * A class for classifier tasks | ||
| * @param modelPathPrefix PathPrefix from where to load the symbol, parameters and synset.txt | ||
| * Example: file://model-dir/resnet-152(containing resnet-152-symbol.json | ||
| * file://model-dir/synset.txt | ||
| * @param inputDescriptors Descriptors defining the input node names, shape, | ||
| * layout and Type parameters | ||
| * @param contexts Device Contexts on which you want to run Inference, defaults to CPU. | ||
| * @param epoch Model epoch to load, defaults to 0. | ||
| */ | ||
| class Classifier(modelPathPrefix: String, | ||
| protected val inputDescriptors: IndexedSeq[DataDesc], | ||
| protected val contexts: Array[Context] = Context.cpu(), | ||
| protected val epoch: Option[Int] = Some(0)) | ||
| extends ClassifierBase { | ||
|
|
||
| private val logger = LoggerFactory.getLogger(classOf[Classifier]) | ||
|
|
||
| protected[infer] val predictor: PredictBase = getPredictor() | ||
|
|
||
| protected[infer] val synsetFilePath = getSynsetFilePath(modelPathPrefix) | ||
|
|
||
| protected[infer] val synset = readSynsetFile(synsetFilePath) | ||
|
|
||
| protected[infer] val handler = MXNetHandler() | ||
|
|
||
| /** | ||
| * Takes a flat arrays as input and returns a List of (Label, tuple) | ||
| * @param input: IndexedSequence one-dimensional array of Floats. | ||
| * @param topK: (Optional) How many top_k(sorting will be based on the last axis) | ||
| * elements to return, if not passed returns unsorted output. | ||
| * @return IndexedSequence of (Label, Score) tuples. | ||
| */ | ||
| override def classify(input: IndexedSeq[Array[Float]], | ||
| topK: Option[Int] = None): IndexedSeq[(String, Float)] = { | ||
|
|
||
| // considering only the first output | ||
| val predictResult = predictor.predict(input)(0) | ||
| var result: IndexedSeq[(String, Float)] = IndexedSeq.empty | ||
|
|
||
| if (topK.isDefined) { | ||
| val sortedIndex = predictResult.zipWithIndex.sortBy(-_._1).map(_._2).take(topK.get) | ||
| result = sortedIndex.map(i => (synset(i), predictResult(i))).toIndexedSeq | ||
| } else { | ||
| result = synset.zip(predictResult).toIndexedSeq | ||
| } | ||
| result | ||
| } | ||
|
|
||
| /** | ||
| * Takes input as NDArrays, useful when you want to perform multiple operations on | ||
| * the input Array or when you want to pass a batch of input. | ||
| * @param input: Indexed Sequence of NDArrays | ||
| * @param topK: (Optional) How many top_k(sorting will be based on the last axis) | ||
| * elements to return, if not passed returns unsorted output. | ||
| * @return Traversable Sequence of (Label, Score) tuple | ||
| */ | ||
| override def classifyWithNDArray(input: IndexedSeq[NDArray], topK: Option[Int] = None) | ||
| : IndexedSeq[IndexedSeq[(String, Float)]] = { | ||
|
|
||
| // considering only the first output | ||
| val predictResultND: NDArray = predictor.predictWithNDArray(input)(0) | ||
|
|
||
| val predictResult: ListBuffer[Array[Float]] = ListBuffer[Array[Float]]() | ||
|
|
||
| // iterating over the individual items(batch size is in axis 0) | ||
| for (i <- 0 until predictResultND.shape(0)) { | ||
| val r = predictResultND.at(i) | ||
| predictResult += r.toArray | ||
| r.dispose() | ||
| } | ||
|
|
||
| var result: ListBuffer[IndexedSeq[(String, Float)]] = | ||
| ListBuffer.empty[IndexedSeq[(String, Float)]] | ||
|
|
||
| if (topK.isDefined) { | ||
| val sortedIndices = predictResult.map(r => | ||
| r.zipWithIndex.sortBy(-_._1).map(_._2).take(topK.get) | ||
| ) | ||
| for (i <- sortedIndices.indices) { | ||
| result += sortedIndices(i).map(sIndx => | ||
| (synset(sIndx), predictResult(i)(sIndx))).toIndexedSeq | ||
| } | ||
| } else { | ||
| for (i <- predictResult.indices) { | ||
| result += synset.zip(predictResult(i)).toIndexedSeq | ||
| } | ||
| } | ||
|
|
||
| handler.execute(predictResultND.dispose()) | ||
|
|
||
| result.toIndexedSeq | ||
| } | ||
|
|
||
| private[infer] def getSynsetFilePath(modelPathPrefix: String): String = { | ||
| val dirPath = modelPathPrefix.substring(0, 1 + modelPathPrefix.lastIndexOf(File.separator)) | ||
| val d = new File(dirPath) | ||
| require(d.exists && d.isDirectory, "directory: %s not found".format(dirPath)) | ||
|
|
||
| val s = new File(dirPath + "synset.txt") | ||
| require(s.exists() && s.isFile, "File synset.txt should exist inside modelPath: %s".format | ||
| (dirPath + "synset.txt")) | ||
|
|
||
| s.getCanonicalPath | ||
| } | ||
|
|
||
| private[infer] def readSynsetFile(synsetFilePath: String): IndexedSeq[String] = { | ||
| val f = io.Source.fromFile(synsetFilePath) | ||
| try { | ||
| f.getLines().toIndexedSeq | ||
| } finally { | ||
| f.close | ||
| } | ||
| } | ||
|
|
||
| private[infer] def getPredictor(): PredictBase = { | ||
| new Predictor(modelPathPrefix, inputDescriptors, contexts, epoch) | ||
| } | ||
|
|
||
| } |
Oops, something went wrong.