PeopleSegNet cannot be used to de inference

swka1043338 · March 28, 2023, 9:51am

Hello. I’ve tried three pretrained models which are from PeopleSegNet NGC official website to model inference. (peoplesegnet_resnet50.etlt, peoplesegnet_resnet50.tlt, and peoplesegnet_resnet50.step-20000.tlt)

However. I got three errors of inference results:

peoplesegnet_resnet50.etlt: ValueError: Model extension needs to be either .engine or .tlt.

image1398×672 80.9 KB
peoplesegnet_resnet50.tlt: AssertionError: The pruned model must be retrained first.

image1402×695 83 KB
peoplesegnet_resnet50.step-20000.tlt:

WARNING:tensorflow:Deprecation warnings have been disabled. Set TF_ENABLE_DEPRECATION_WARNINGS=1 to re-enable them.
/usr/local/lib/python3.6/dist-packages/requests/__init__.py:91: RequestsDependencyWarning: urllib3 (1.26.5) or chardet (3.0.4) doesn't match a supported version!
  RequestsDependencyWarning)
Using TensorFlow backend.
2023-03-28 09:38:42,813 [INFO] root: Starting MaskRCNN inference.
Label file does not exist. Skipping...
2023-03-28 09:38:42,813 [INFO] iva.mask_rcnn.utils.spec_loader: Loading specification from /workspace/tao-experiments/maskrcnn_retrain_resnet50.txt
INFO:tensorflow:Using config: {'_model_dir': '/tmp/tmpuk9wazjk', '_tf_random_seed': 123, '_save_summary_steps': None, '_save_checkpoints_steps': None, '_save_checkpoints_secs': None, '_session_config': gpu_options {
  allow_growth: true
  force_gpu_compatible: true
}
allow_soft_placement: true
graph_options {
  rewrite_options {
    meta_optimizer_iterations: TWO
  }
}
, '_keep_checkpoint_max': 20, '_keep_checkpoint_every_n_hours': None, '_log_step_count_steps': None, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_experimental_max_worker_delay_secs': None, '_session_creation_timeout_secs': 7200, '_service': None, '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x7f89c34f7080>, '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1}
2023-03-28 09:38:42,816 [INFO] tensorflow: Using config: {'_model_dir': '/tmp/tmpuk9wazjk', '_tf_random_seed': 123, '_save_summary_steps': None, '_save_checkpoints_steps': None, '_save_checkpoints_secs': None, '_session_config': gpu_options {
  allow_growth: true
  force_gpu_compatible: true
}
allow_soft_placement: true
graph_options {
  rewrite_options {
    meta_optimizer_iterations: TWO
  }
}
, '_keep_checkpoint_max': 20, '_keep_checkpoint_every_n_hours': None, '_log_step_count_steps': None, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_experimental_max_worker_delay_secs': None, '_session_creation_timeout_secs': 7200, '_service': None, '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x7f89c34f7080>, '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1}
[MaskRCNN] INFO    : Running inference...
[MaskRCNN] INFO    : Loading weights from /workspace/tao-experiments/peoplesegnet_resnet50.step-20000.tlt
2023-03-28 09:38:45,094 [INFO] root: The last checkpoint file is not saved properly.                 Please delete it and rerun the script.
Traceback (most recent call last):
  File "<frozen iva.mask_rcnn.executer.distributed_executer>", line 352, in extract_ckpt
  File "/usr/lib/python3.6/zipfile.py", line 1131, in __init__
    self._RealGetContents()
  File "/usr/lib/python3.6/zipfile.py", line 1198, in _RealGetContents
    raise BadZipFile("File is not a zip file")
zipfile.BadZipFile: File is not a zip file

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "</usr/local/lib/python3.6/dist-packages/iva/mask_rcnn/scripts/inference.py>", line 3, in <module>
  File "<frozen iva.mask_rcnn.scripts.inference>", line 390, in <module>
  File "<frozen iva.mask_rcnn.scripts.inference>", line 378, in <module>
  File "<frozen iva.mask_rcnn.scripts.inference>", line 365, in main
  File "<frozen iva.mask_rcnn.scripts.inference>", line 311, in infer
  File "<frozen iva.mask_rcnn.executer.distributed_executer>", line 503, in infer
  File "<frozen iva.mask_rcnn.executer.distributed_executer>", line 357, in extract_ckpt
OSError: The last checkpoint file is not saved properly.                 Please delete it and rerun the script.
Telemetry data couldn't be sent, but the command ran successfully.
[WARNING]: <urlopen error [Errno -2] Name or service not known>
Execution status: FAIL

This is my command:

docker run -it --rm -v /home/ubuntu/tao_test_2023/tensorflow_train/model_zoo/peoplesegnet:/workspace/tao-experiments 
                    nvcr.io/nvidia/tao/tao-toolkit:4.0.1-tf1.15.5 
                    mask_rcnn inference 
                    -i /workspace/tao-experiments/data -o /workspace/tao-experiments/ 
                    -e /workspace/tao-experiments/maskrcnn_retrain_resnet50.txt 
                    -m /workspace/tao-experiments/peoplesegnet_resnet50.tlt 
                    -l /workspace/tao-experiments/coco_labels.txt -t 0.5 
                    -k nvidia_tao 
                    --include_mask

How should I do generate inference result correctly ?

Morganh · March 29, 2023, 1:59am

When running inference with “mask_rcnn inference”, it is using .tlt file or .engine file.

Can you share training spec file and command?

swka1043338 · April 2, 2023, 4:04am

Here is my spec file used for inference. Which parts should I edit to deal with the problem OSError: The last checkpoint file is not saved properly ?

seed: 123
use_amp: False
warmup_steps: 10000
learning_rate_steps: "[100000, 150000, 200000]"
learning_rate_decay_levels: "[0.1, 0.02, 0.01]"
total_steps: 250000
train_batch_size: 2
eval_batch_size: 4
num_steps_per_eval: 5000
momentum: 0.9
l2_weight_decay: 0.00004
warmup_learning_rate: 0.0001
init_learning_rate: 0.005
num_examples_per_epoch: 118288
pruned_model_path: "/workspace/tao-experiments/mask_rcnn/peoplesegnet_resnet50.step-20000.tlt"

data_config{
    image_size: "(576, 960)"
    augment_input_data: True
    eval_samples: 500
    training_file_pattern: "/workspace/tao-experiments/data/train*.tfrecord"
    validation_file_pattern: "/workspace/tao-experiments/data/val*.tfrecord"
    val_json_file: "/workspace/tao-experiments/data/raw-data/annotations/instances_val2017.json"

    # dataset specific parameters
    num_classes: 91
    skip_crowd_during_training: True
}

maskrcnn_config {
    nlayers: 50
    arch: "resnet"
    freeze_bn: True
    freeze_blocks: "[0,1]"
    gt_mask_size: 112
        
    # Region Proposal Network
    rpn_positive_overlap: 0.7
    rpn_negative_overlap: 0.3
    rpn_batch_size_per_im: 256
    rpn_fg_fraction: 0.5
    rpn_min_size: 0.

    # Proposal layer.
    batch_size_per_im: 512
    fg_fraction: 0.25
    fg_thresh: 0.5
    bg_thresh_hi: 0.5
    bg_thresh_lo: 0.

    # Faster-RCNN heads.
    fast_rcnn_mlp_head_dim: 1024
    bbox_reg_weights: "(10., 10., 5., 5.)"

    # Mask-RCNN heads.
    include_mask: True
    mrcnn_resolution: 28

    # training
    train_rpn_pre_nms_topn: 2000
    train_rpn_post_nms_topn: 1000
    train_rpn_nms_threshold: 0.7

    # evaluation
    test_detections_per_image: 100
    test_nms: 0.5
    test_rpn_pre_nms_topn: 1000
    test_rpn_post_nms_topn: 1000
    test_rpn_nms_thresh: 0.7

    # model architecture
    min_level: 2
    max_level: 6
    num_scales: 1
    aspect_ratios: "[(1.0, 1.0), (1.4, 0.7), (0.7, 1.4)]"
    anchor_scale: 8

    # localization loss
    rpn_box_loss_weight: 1.0
    fast_rcnn_box_loss_weight: 1.0
    mrcnn_weight_loss_mask: 1.0
}

Morganh · April 2, 2023, 4:02pm

There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks

Please refer to the command mentioned in MaskRCNN - NVIDIA Docs
You can set the model in “-m”.

tao mask_rcnn inference [-h] -i <input directory>
                             -o <output annotated image directory>
                             -e <experiment spec file>
                             -m <model file>
                             -k <key>

system · May 2, 2023, 4:11am

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.

Topic		Replies	Views
No detections after training PeopleNet using custom labeled data TAO Toolkit	7	867	October 12, 2021
Trouble using trt-infer on peoplenet pretrained model TAO Toolkit	11	790	October 12, 2021
Basic questions about transfer learning with TAO Toolkit TAO Toolkit	2	458	January 12, 2023
Run PeopleNet with tensorrt TAO Toolkit	35	9788	August 10, 2021
Accelerating Peoplnet with tlt for jetson nano TAO Toolkit	19	2406	October 12, 2021
Run peoplenet on python virtual environment without jetson TAO Toolkit tensorrt	29	1124	January 20, 2024
PeopleNet. Coverage output is always zero TAO Toolkit	12	1801	October 12, 2021
Tao inference ValueError: could not broadcast input array from shape (4700160) into shape (1566720) TAO Toolkit	9	641	December 9, 2022
Errors while running inference on Peoplenet Model (no training) TAO Toolkit	3	451	December 28, 2021
Peoplenet unpruned model evaluation TAO Toolkit	8	690	October 12, 2021

PeopleSegNet cannot be used to de inference

Related topics