From b3a14f46815aa9eb8d637f2a7a9c4d7cdda7f3df Mon Sep 17 00:00:00 2001
From: Kaipeng Deng <dengkaipeng@baidu.com>
Date: Sat, 14 Dec 2019 13:36:26 +0800
Subject: [PATCH] fix pointrcnn (#4073)

---
 PaddleCV/Paddle3D/PointRCNN/README.md                 |  3 ++-
 PaddleCV/Paddle3D/PointRCNN/data/kitti_rcnn_reader.py |  9 +++++++++
 PaddleCV/Paddle3D/PointRCNN/train.py                  | 10 +++++++++-
 3 files changed, 20 insertions(+), 2 deletions(-)

diff --git a/PaddleCV/Paddle3D/PointRCNN/README.md b/PaddleCV/Paddle3D/PointRCNN/README.md
index 4c85db54..842cb86d 100644
--- a/PaddleCV/Paddle3D/PointRCNN/README.md
+++ b/PaddleCV/Paddle3D/PointRCNN/README.md
@@ -232,7 +232,8 @@ python train.py --cfg=./cfgs/default.yml \
                 --epoch=30 \
                 --save_dir=checkpoints \
                 --rcnn_training_roi_dir=output/detections/data \
-                --rcnn_training_feature_dir=output/features
+                --rcnn_training_feature_dir=output/features \
+                --set TRAIN.SPLIT train_aug
 ```
 
 RCNN模型训练权重默认保存在`checkpoints/rcnn`目录下，可通过`--save_dir`参数指定。
diff --git a/PaddleCV/Paddle3D/PointRCNN/data/kitti_rcnn_reader.py b/PaddleCV/Paddle3D/PointRCNN/data/kitti_rcnn_reader.py
index 811a20b2..57367d2c 100644
--- a/PaddleCV/Paddle3D/PointRCNN/data/kitti_rcnn_reader.py
+++ b/PaddleCV/Paddle3D/PointRCNN/data/kitti_rcnn_reader.py
@@ -20,6 +20,7 @@ from __future__ import division
 from __future__ import print_function
 
 import os
+import signal
 import logging
 import multiprocessing
 import numpy as np
@@ -1182,3 +1183,11 @@ class KittiRCNNReader(KittiDataset):
 
         return reader
 
+
+def _term_reader(signum, frame):
+    logger.info('pid {} terminated, terminate reader process '
+                'group {}...'.format(os.getpid(), os.getpgrp()))
+    os.killpg(os.getpgid(os.getpid()), signal.SIGKILL)
+
+signal.signal(signal.SIGINT, _term_reader)
+
diff --git a/PaddleCV/Paddle3D/PointRCNN/train.py b/PaddleCV/Paddle3D/PointRCNN/train.py
index b7a39ca4..41a6f098 100644
--- a/PaddleCV/Paddle3D/PointRCNN/train.py
+++ b/PaddleCV/Paddle3D/PointRCNN/train.py
@@ -98,6 +98,11 @@ def parse_args():
         type=str,
         default=None,
 	help='specify the saved features for rcnn training when using rcnn_offline mode')
+    parser.add_argument(
+        '--worker_num',
+        type=int,
+        default=16,
+	help='multiprocess reader process num, default 16')
     parser.add_argument(
         '--log_interval',
         type=int,
@@ -206,7 +211,10 @@ def train():
         fluid.io.save_persistables(exe, path, prog)
 
     # get reader
-    train_reader = kitti_rcnn_reader.get_multiprocess_reader(args.batch_size, train_feeds, drop_last=True)
+    train_reader = kitti_rcnn_reader.get_multiprocess_reader(args.batch_size,
+                                                             train_feeds,
+                                                             proc_num=args.worker_num,
+                                                             drop_last=True)
     train_pyreader.decorate_sample_list_generator(train_reader, place)
 
     train_stat = Stat()
-- 
GitLab