Scala Infer ObjectDector crashes on models which don't return

[andrewfayres]

Description

The Scala Infer API's ObjectDetector will crash when provided with a model which doesn't return coordinates for bounding boxes. We should handle this situation more gracefully. Either return an error message stating that the model doesn't have the required outputs or return the results without the bounding boxes.

Error Message:

$ ./bin/run_det.sh
SLF4J: Failed to load class "org.slf4j.impl.StaticLoggerBinder".
SLF4J: Defaulting to no-operation (NOP) logger implementation
SLF4J: See http:https://www.slf4j.org/codes.html#StaticLoggerBinder for further details.
[22:18:13] src/nnvm/legacy_json_util.cc:209: Loading symbol saved by previous version v1.3.1. Attempting to upgrade...
[22:18:13] src/nnvm/legacy_json_util.cc:217: Symbol successfully upgraded!
[22:18:17] src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:97: Running performance tests to find the best convolution algorithm, this can take a while... (setting env variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to disable)
Exception in thread "main" java.lang.ArrayIndexOutOfBoundsException: 1
    at org.apache.mxnet.infer.ObjectDetector$$anonfun$2.apply(ObjectDetector.scala:132)
    at org.apache.mxnet.infer.ObjectDetector$$anonfun$2.apply(ObjectDetector.scala:132)
    at scala.math.Ordering$$anon$5.compare(Ordering.scala:122)
    at java.util.TimSort.countRunAndMakeAscending(TimSort.java:355)
    at java.util.TimSort.sort(TimSort.java:234)
    at java.util.Arrays.sort(Arrays.java:1438)
    at scala.collection.SeqLike$class.sorted(SeqLike.scala:648)
    at scala.collection.mutable.ArrayOps$ofRef.sorted(ArrayOps.scala:186)
    at scala.collection.SeqLike$class.sortBy(SeqLike.scala:623)
    at scala.collection.mutable.ArrayOps$ofRef.sortBy(ArrayOps.scala:186)
    at org.apache.mxnet.infer.ObjectDetector.sortAndReformat(ObjectDetector.scala:132)
    at org.apache.mxnet.infer.ObjectDetector$$anonfun$objectDetectWithNDArray$1.apply$mcVI$sp(ObjectDetector.scala:106)
    at org.apache.mxnet.infer.ObjectDetector$$anonfun$objectDetectWithNDArray$1.apply(ObjectDetector.scala:104)
    at org.apache.mxnet.infer.ObjectDetector$$anonfun$objectDetectWithNDArray$1.apply(ObjectDetector.scala:104)
    at scala.collection.parallel.mutable.ParArray$ParArrayIterator.foreach_quick(ParArray.scala:143)
    at scala.collection.parallel.mutable.ParArray$ParArrayIterator.foreach(ParArray.scala:136)
    at scala.collection.parallel.ParIterableLike$Foreach.leaf(ParIterableLike.scala:972)
    at scala.collection.parallel.Task$$anonfun$tryLeaf$1.apply$mcV$sp(Tasks.scala:49)
    at scala.collection.parallel.Task$$anonfun$tryLeaf$1.apply(Tasks.scala:48)
    at scala.collection.parallel.Task$$anonfun$tryLeaf$1.apply(Tasks.scala:48)
    at scala.collection.parallel.Task$class.tryLeaf(Tasks.scala:51)
    at scala.collection.parallel.ParIterableLike$Foreach.tryLeaf(ParIterableLike.scala:969)
    at scala.collection.parallel.AdaptiveWorkStealingTasks$WrappedTask$class.compute(Tasks.scala:152)
    at scala.collection.parallel.AdaptiveWorkStealingForkJoinTasks$WrappedTask.compute(Tasks.scala:443)
    at scala.concurrent.forkjoin.RecursiveAction.exec(RecursiveAction.java:160)
    at scala.concurrent.forkjoin.ForkJoinTask.doExec(ForkJoinTask.java:260)
    at scala.concurrent.forkjoin.ForkJoinPool$WorkQueue.runTask(ForkJoinPool.java:1339)
    at scala.concurrent.forkjoin.ForkJoinPool.runWorker(ForkJoinPool.java:1979)
    at scala.concurrent.forkjoin.ForkJoinWorkerThread.run(ForkJoinWorkerThread.java:107)

Minimum reproducible example

    public static void main(String[] args) {
        String modelPathPrefix = "./resnet-152";
        String inputImagePath = "./cat.jpg";

        List<Context> context = new ArrayList<Context>();
        context.add(Context.gpu());

        // Prepare the model                                                                                                                          
        List<DataDesc> inputDesc = new ArrayList<>();
        Shape inputShape = new Shape(new int[]{1, 3, 224, 224});
        inputDesc.add(new DataDesc("data", inputShape, DType.Float32(), "NCHW"));
        ObjectDetector objDet = new ObjectDetector(modelPathPrefix, inputDesc, context, 0);

        // predict                                                                                                                                    
        int iter = 100;
        ArrayList<Double> data = new ArrayList<Double>(iter);
        for(int i=0; i<iter; i++) {
            // Prepare data                                                                                                                           
            BufferedImage img = ObjectDetector.loadImageFromFile(inputImagePath);

            long start = System.nanoTime();
            List<List<ObjectDetectorOutput>> result = objDet.imageObjectDetect(img,3);
            long stop = System.nanoTime();

            double elapsed = (stop-start)/1000000.0;
            data.add(elapsed);
        }

Steps to reproduce

(Paste the commands you ran that produced the error.)

1. Download resnet-152 model
2. Run above code calling ObjectDetector (above code is Java and call the Java API but error is in the Scala layer)

[mxnet-label-bot]

Hey, this is the MXNet Label Bot.
Thank you for submitting the issue! I will try and suggest some labels so that the appropriate MXNet community members can help resolve it.
Here are my recommended labels: Scala, Bug

[andrewfayres]

@mxnet-label-bot add [Bug, Scala]