DA-RNN icon indicating copy to clipboard operation
DA-RNN copied to clipboard
verified publisher icon yuxng

Error when test the trained model

Open gaochuan2017 opened this issue 7 years ago • 0 comments

I'm sure I have installed correct version of dependencies ,including Eigen,Sophus,Pangolin and so on.I'm not meeting Errors when "sh make.sh",I also succeed in installing KinectFusion ,that part is OK.But When I download and test the trained model Error occurs.It said "Framebuffer with requested attributes not available. Using available framebuffer. You may see visual artifacts.created window" . The details are as follows:

:~/DA-RNN-master$ ./experiments/scripts/rgbd_scene_multi_rgbd_test.sh 0

  • set -e
  • export PYTHONUNBUFFERED=True
  • PYTHONUNBUFFERED=True
  • export CUDA_VISIBLE_DEVICES=0
  • CUDA_VISIBLE_DEVICES=0 ++ date +%Y-%m-%d_%H-%M-%S
  • LOG=experiments/logs/rgbd_scene_multi_rgbd_test.txt.2019-01-11_15-03-36
  • exec ++ tee -a experiments/logs/rgbd_scene_multi_rgbd_test.txt.2019-01-11_15-03-36
  • echo Logging output to experiments/logs/rgbd_scene_multi_rgbd_test.txt.2019-01-11_15-03-36 Logging output to experiments/logs/rgbd_scene_multi_rgbd_test.txt.2019-01-11_15-03-36
  • '[' -f /home/gaochuan/DA-RNN-master/output/rgbd_scene/rgbd_scene_val/vgg16_fcn_rgbd_multi_frame_rgbd_scene_iter_40000/segmentations.pkl ']'
  • ./tools/test_net.py --gpu 0 --network vgg16 --model data/fcn_models/rgbd_scene/vgg16_fcn_rgbd_multi_frame_rgbd_scene_iter_40000.ckpt --imdb rgbd_scene_val --cfg experiments/cfgs/rgbd_scene_multi_rgbd.yml --rig data/RGBDScene/camera.json --kfusion 1 registered Linear registered Linear registered Poly3 registered Poly3 shapenet_scene_train shapenet_scene_val shapenet_single_train shapenet_single_val gmu_scene_train gmu_scene_val rgbd_scene_train rgbd_scene_val rgbd_scene_trainval lov_train lov_val Called with args: Namespace(cfg_file='experiments/cfgs/rgbd_scene_multi_rgbd.yml', gpu_id=0, imdb_name='rgbd_scene_val', kfusion=True, model='data/fcn_models/rgbd_scene/vgg16_fcn_rgbd_multi_frame_rgbd_scene_iter_40000.ckpt', network_name='vgg16', pretrained_model=None, rig_name='data/RGBDScene/camera.json', wait=True) Using config: {'EPS': 1e-14, 'EXP_DIR': 'rgbd_scene', 'FLIP_X': False, 'GPU_ID': 0, 'INPUT': 'RGBD', 'NETWORK': 'VGG16', 'PIXEL_MEANS': array([[[102.9801, 115.9465, 122.7717]]]), 'RNG_SEED': 3, 'ROOT_DIR': '/home/gaochuan/DA-RNN-master', 'TEST': {'GRID_SIZE': 512, 'RANSAC': False, 'SCALES_BASE': [1.0], 'SINGLE_FRAME': False, 'VERTEX_REG': False, 'VISUALIZE': False}, 'TRAIN': {'CHROMATIC': True, 'DISPLAY': 20, 'GAMMA': 0.1, 'GRID_SIZE': 512, 'IMS_PER_BATCH': 1, 'LEARNING_RATE': 0.0001, 'MOMENTUM': 0.9, 'NUM_CLASSES': 10, 'NUM_STEPS': 3, 'NUM_UNITS': 64, 'SCALES_BASE': [1.0], 'SINGLE_FRAME': False, 'SNAPSHOT_INFIX': 'rgbd_scene', 'SNAPSHOT_ITERS': 10000, 'SNAPSHOT_PREFIX': 'vgg16_fcn_rgbd_multi_frame', 'STEPSIZE': 30000, 'TRAINABLE': True, 'USE_FLIPPED': False, 'VERTEX_REG': False, 'VERTEX_W': 10.0, 'VISUALIZE': False}} /gpu:0 Tensor("unstack:0", dtype=float32) Tensor("unstack_1:0", dtype=float32) Tensor("conv5_3/conv5_3:0", shape=(?, ?, ?, 512), dtype=float32) Tensor("conv5_3_p/conv5_3_p:0", shape=(?, ?, ?, 512), dtype=float32) Tensor("conv4_3/conv4_3:0", shape=(?, ?, ?, 512), dtype=float32) Tensor("conv4_3_p/conv4_3_p:0", shape=(?, ?, ?, 512), dtype=float32) Tensor("score_conv4/score_conv4:0", shape=(?, ?, ?, 64), dtype=float32) Tensor("upscore_conv5_1:0", shape=(?, ?, ?, 64), dtype=float32) Tensor("fifo_queue_Dequeue:5", dtype=float32) Tensor("fifo_queue_Dequeue:6", dtype=float32) Tensor("fifo_queue_Dequeue:7", dtype=float32) Tensor("unstack_3:0", dtype=float32) Tensor("unstack_4:0", dtype=float32) Tensor("upscore_1:0", shape=(?, ?, ?, 64), dtype=float32) Computeflow(top_data=<tf.Tensor 'flow:0' shape= dtype=float32>, top_weights=<tf.Tensor 'flow:1' shape= dtype=float32>, top_points=<tf.Tensor 'flow:2' shape= dtype=float32>) Tensor("score/score:0", shape=(?, ?, ?, 10), dtype=float32) Use network vgg16 in training 2019-01-11 15:03:37.225146: I tensorflow/core/platform/cpu_feature_guard.cc:137] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX AVX2 FMA 2019-01-11 15:03:37.339921: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:892] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero 2019-01-11 15:03:37.340290: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1030] Found device 0 with properties: name: GeForce RTX 2080 major: 7 minor: 5 memoryClockRate(GHz): 1.71 pciBusID: 0000:01:00.0 totalMemory: 7.76GiB freeMemory: 7.48GiB 2019-01-11 15:03:37.340301: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1120] Creating TensorFlow device (/device:GPU:0) -> (device: 0, name: GeForce RTX 2080, pci bus id: 0000:01:00.0, compute capability: 7.5) Loading model weights from data/fcn_models/rgbd_scene/vgg16_fcn_rgbd_multi_frame_rgbd_scene_iter_40000.ckpt rgbd_scene_val aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa {"down":[0,1,0],"forward":[0,0,1],"height":480,"param_names":["fu","fv","u0","v0","k1","k2","k3"],"params":[570.29999999999995,570.29999999999995,320,240,0,0,0],"right":[1,0,0],"serialno":"34178534347","type":"Poly3","width":640} "Poly3" params: 570.29999999999995 570.29999999999995 320 240 0 0 0
    pose: [ [ 1, 0, 0, 0 ], [ 0, 1, 0, 0 ], [ 0, 0, 1, 0 ] ]

{"down":[0,1,0],"forward":[0,0,1],"height":480,"param_names":["fu","fv","u0","v0","k1","k2","k3"],"params":[570.29999999999995,570.29999999999995,320,240,0,0,0],"right":[1,0,0],"serialno":"34178534347","type":"Poly3","width":640} "Poly3" params: 570.29999999999995 570.29999999999995 320 240 0 0 0
pose: [ [ 1, 0, 0, 0 ], [ 0, 1, 0, 0 ], [ 0, 0, 1, 0 ] ]

T_dc: 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 Framebuffer with requested attributes not available. Using available framebuffer. You may see visual artifacts.created window ./experiments/scripts/rgbd_scene_multi_rgbd_test.sh: 行 25: 15989 段错误 (核心已转储) ./tools/test_net.py --gpu 0 --network vgg16 --model data/fcn_models/rgbd_scene/vgg16_fcn_rgbd_multi_frame_rgbd_scene_iter_40000.ckpt --imdb rgbd_scene_val --cfg experiments/cfgs/rgbd_scene_multi_rgbd.yml --rig data/RGBDScene/camera.json --kfusion 1

"段错误(核心转存储)" means “segmentation fault (core dumped)”. I check the code and I have found in fact Error is at the line 309 of test.py : if is_kfusion: KF = kfusion.PyKinectFusion(rig_filename) But the KinectFusion has been installed correctly so I don't know what causes the Error. Probable Reason may be: 1.I use RTX 2080 ,with Cuda 8.0 ,Cudnn 6.0.The device is not compatible. 2.I'm using a new computer,I have not installed OpenCV and other libraries in my computer yet.

@yuxng any suggestions? I'm vert interested in this superb project and thanks a lot .

gaochuan2017 avatar Jan 11 '19 07:01 gaochuan2017