update scripts

object_detection/README.md

@@ -1,68 +1,87 @@
 # 1. Problem
+
 Object detection and segmentation. Metrics are mask and box mAP.
 
 # 2. Directions
 
 ### Steps to configure machine
 
 1. Checkout the MLPerf repository
+
 ```
 mkdir -p mlperf
 cd mlperf
 git clone https://github.com/mlperf/training.git
 ```
+
 2. Install CUDA and Docker
+
 ```
 source training/install_cuda_docker.sh
 ```
+
 3. Build the docker image for the object detection task
+
 ```
 cd training/object_detection/
 nvidia-docker build . -t mlperf/object_detection
 ```
 
 ### Steps to download data
+
 ```
 # From training/object_detection/
 source download_dataset.sh
 ```
 
-### Steps to run benchmark.
+### Steps to run benchmark
+
 ```
 nvidia-docker run -v .:/workspace -t -i --rm --ipc=host mlperf/object_detection \
     "cd mlperf/training/object_detection && ./run_and_time.sh"
 ```
 
 # 3. Dataset/Environment
+
 ### Publication/Attribution
+
 Microsoft COCO: Common Objects in Context
 
 ### Data preprocessing
+
 Only horizontal flips are allowed.
 
 ### Training and test data separation
+
 As provided by MS-COCO (2017 version).
 
 ### Training data order
+
 Randomly.
 
 ### Test data order
+
 Any order.
 
 # 4. Model
+
 ### Publication/Attribution
+
 He, Kaiming, et al. "Mask r-cnn." Computer Vision (ICCV), 2017 IEEE International Conference on.
 IEEE, 2017.
 
 We use a version of Mask R-CNN with a ResNet-50 backbone.
 
 ### List of layers
+
 Running the timing script will display a list of layers.
 
 ### Weight and bias initialization
+
 The ResNet-50 base must be loaded from the provided weights. They may be quantized.
 
 ### Loss function
+
 Multi-task loss (classification, box, mask). Described in the Mask R-CNN paper.
 
 Classification: Smooth L1 loss
@@ -72,17 +91,23 @@ Box: Log loss for true class.
 Mask: per-pixel sigmoid, average binary cross-entropy loss.
 
 ### Optimizer
+
 Momentum SGD. Weight decay of 0.0001, momentum of 0.9.
 
 # 5. Quality
+
 ### Quality metric
+
 As Mask R-CNN can provide both boxes and masks, we evaluate on both box and mask mAP.
 
 ### Quality target
+
 Box mAP of 0.377, mask mAP of 0.339
 
 ### Evaluation frequency
+
 Once per epoch, 118k.
 
 ### Evaluation thoroughness
+
 Evaluate over the entire validation set. Use the official COCO API to compute mAP.

object_detection/download_dataset.sh

@@ -11,7 +11,6 @@ while [ $# -gt 0 ]; do
   shift
 done
 
-
 # Get COCO 2017 data sets
 echo "Downloaading to folder: $DATA_ROOT_DIR"
 mkdir -p $DATA_ROOT_DIR
@@ -25,30 +24,41 @@ tar -xzf coco_annotations_minival.tgz &>/dev/null
 echo "Downloading annotations_trainval2017.zip:"
 curl -O http://images.cocodataset.org/annotations/annotations_trainval2017.zip
 echo "Extracting annotations_trainval2017.zip:"
-n_files=`unzip -l  annotations_trainval2017.zip| grep .json | wc -l`
-unzip annotations_trainval2017.zip  | { I=-1; while read; do printf "Progress: $((++I*100/$n_files))%%\r"; done; echo ""; }
+n_files=$(unzip -l annotations_trainval2017.zip | grep .json | wc -l)
+unzip annotations_trainval2017.zip | {
+  I=-1
+  while read; do printf "Progress: $((++I * 100 / $n_files))%%\r"; done
+  echo ""
+}
 
 echo "Downloading val2017.zip:"
 curl -O http://images.cocodataset.org/zips/val2017.zip
 echo "Extracting val2017.zip:"
-n_files=`unzip -l  val2017.zip| grep .jpg | wc -l`
-unzip val2017.zip  | { I=-1; while read; do printf "Progress: $((++I*100/$n_files))%%\r"; done; echo ""; }
+n_files=$(unzip -l val2017.zip | grep .jpg | wc -l)
+unzip val2017.zip | {
+  I=-1
+  while read; do printf "Progress: $((++I * 100 / $n_files))%%\r"; done
+  echo ""
+}
 
 echo "Downloading train2017.zip:"
 curl -O http://images.cocodataset.org/zips/train2017.zip
 echo "Extracting train2017.zip:"
-n_files=`unzip -l  train2017.zip| grep .jpg | wc -l`
-unzip train2017.zip  | { I=-1; while read; do printf "Progress: $((++I*100/$n_files))%%\r"; done; echo ""; }
+n_files=$(unzip -l train2017.zip | grep .jpg | wc -l)
+unzip train2017.zip | {
+  I=-1
+  while read; do printf "Progress: $((++I * 100 / $n_files))%%\r"; done
+  echo ""
+}
 
 # MD5 verification
 echo "Running MD5 verification ... this might take a while"
-checkMD5 () {
-    if [ $(pv -f $1| md5sum | cut -d' ' -f1) = $2 ];
-    then
-        echo "$1 MD5 is valid"
-    else 
-        echo "*ERROR* $1 MD5 is NOT valid"
-    fi
+checkMD5() {
+  if [ $(pv -f $1 | md5sum | cut -d' ' -f1) = $2 ]; then
+    echo "$1 MD5 is valid"
+  else
+    echo "*ERROR* $1 MD5 is NOT valid"
+  fi
 }
 
 echo "validating annotations_trainval2017.zip:"

object_detection/download_demo.sh

@@ -11,7 +11,6 @@ while [ $# -gt 0 ]; do
   shift
 done
 
-
 echo "Downloaading demo to folder: $DATA_ROOT_DIR"
 mkdir -p $DATA_ROOT_DIR
 pushd $DATA_ROOT_DIR
@@ -24,13 +23,12 @@ echo "Done!"
 
 # MD5 verification
 echo "Running MD5 verification ..."
-checkMD5 () {
-    if [ $(pv -f $1| md5sum | cut -d' ' -f1) = $2 ];
-    then
-        echo "$1 MD5 is valid"
-    else 
-        echo "*ERROR* $1 MD5 is NOT valid"
-    fi
+checkMD5() {
+  if [ $(pv -f $1 | md5sum | cut -d' ' -f1) = $2 ]; then
+    echo "$1 MD5 is valid"
+  else
+    echo "*ERROR* $1 MD5 is NOT valid"
+  fi
 }
 
 echo "validating demo_data.zip:"

object_detection/run_and_time.sh

@@ -28,18 +28,18 @@ SOLVER_MAX_ITER="${SOLVER_MAX_ITER:-40000}"
 
 #Link input data paths
 if [ "$DATA_DIR" != "$DATA_ROOT_TARGET" ]; then
-       mkdir -p $DATA_ROOT_TARGET
-       ln -s $DATA_DIR/annotations $DATA_ROOT_TARGET/annotations
-       ln -s $DATA_DIR/train2017 $DATA_ROOT_TARGET/train2017
-       ln -s $DATA_DIR/test2017 $DATA_ROOT_TARGET/test2017
-       ln -s $DATA_DIR/val2017 $DATA_ROOT_TARGET/val2017
+  mkdir -p $DATA_ROOT_TARGET
+  ln -s $DATA_DIR/annotations $DATA_ROOT_TARGET/annotations
+  ln -s $DATA_DIR/train2017 $DATA_ROOT_TARGET/train2017
+  ln -s $DATA_DIR/test2017 $DATA_ROOT_TARGET/test2017
+  ln -s $DATA_DIR/val2017 $DATA_ROOT_TARGET/val2017
 fi
 
 pwd
 
 # Single GPU training
 time python tools/train_mlperf.py --config-file "configs/e2e_mask_rcnn_R_50_FPN_1x.yaml" \
-       SOLVER.IMS_PER_BATCH 2 TEST.IMS_PER_BATCH 1 SOLVER.MAX_ITER 720000 \
-       SOLVER.STEPS "(480000, 640000)" SOLVER.BASE_LR 0.0025 OUTPUT_DIR $OUTPUT_DIR
-       
+  SOLVER.IMS_PER_BATCH 2 TEST.IMS_PER_BATCH 1 SOLVER.MAX_ITER 720000 \
+  SOLVER.STEPS "(480000, 640000)" SOLVER.BASE_LR 0.0025 OUTPUT_DIR $OUTPUT_DIR
+
 popd

object_detection/run_demo.sh

@@ -28,17 +28,17 @@ SOLVER_MAX_ITER="${SOLVER_MAX_ITER:-40000}"
 
 #Link input data paths
 if [ "$DATA_DIR" != "$DATA_ROOT_TARGET" ]; then
-       mkdir -p $DATA_ROOT_TARGET
-       ln -s $DATA_DIR/annotations $DATA_ROOT_TARGET/annotations
-       ln -s $DATA_DIR/train2017 $DATA_ROOT_TARGET/train2017
-       ln -s $DATA_DIR/val2017 $DATA_ROOT_TARGET/val2017
+  mkdir -p $DATA_ROOT_TARGET
+  ln -s $DATA_DIR/annotations $DATA_ROOT_TARGET/annotations
+  ln -s $DATA_DIR/train2017 $DATA_ROOT_TARGET/train2017
+  ln -s $DATA_DIR/val2017 $DATA_ROOT_TARGET/val2017
 fi
 
 pwd
 
 # Single GPU training
 time python tools/train_mlperf.py --config-file "configs/e2e_mask_rcnn_R_50_FPN_1x.yaml" \
-       SOLVER.IMS_PER_BATCH 2 TEST.IMS_PER_BATCH 1 SOLVER.MAX_ITER 720 \
-       SOLVER.STEPS "(480, 640)" SOLVER.BASE_LR 0.0025 OUTPUT_DIR $OUTPUT_DIR
-       
+  SOLVER.IMS_PER_BATCH 2 TEST.IMS_PER_BATCH 1 SOLVER.MAX_ITER 720 \
+  SOLVER.STEPS "(480, 640)" SOLVER.BASE_LR 0.0025 OUTPUT_DIR $OUTPUT_DIR
+
 popd