Models: ValueError: second_stage_batch_size should be no greater than first_stage_max_proposals.

[mattryles]

• What is the top-level directory of the model you are using:
  object_detection
• Have I written custom code (as opposed to using a stock example script provided in TensorFlow):
  No
• OS Platform and Distribution (e.g., Linux Ubuntu 16.04):
  Ubuntu 16.04
• TensorFlow installed from (source or binary):
  Binary
• TensorFlow version (use command below):
  ('v1.3.0-rc2-20-g0787eee', '1.3.0')
• Bazel version (if compiling from source):
• CUDA/cuDNN version:
  8/6
• GPU model and memory:
  1070 8105MiB Driver 384.90
• Exact command to reproduce:
  train.py

Hi,

Using the default config provided in the samples dir I get this error:

ValueError: second_stage_batch_size should be no greater than first_stage_max_proposals.

Edit: Using the new faster_rcnn_nas pre trained model.

Thanks

Matt

[mattryles]

Adding
second_stage_batch_size: 49
to the config fixed this issue

[wpq3142]

@mattryles “Adding second_stage_batch_size: 49 to the config fixed this issue”

Adding   to the config what position？

}
first_stage_nms_score_threshold: 0.0
first_stage_nms_iou_threshold: 0.7
first_stage_max_proposals: 50
second_stage_batch_size: 49 ----------------------here?
first_stage_localization_loss_weight: 2.0
first_stage_objectness_loss_weight: 1.0
initial_crop_size: 17
maxpool_kernel_size: 1
maxpool_stride: 1
second_stage_box_predictor {
mask_rcnn_box_predictor {

[mattryles]

after second_stage_classification_loss_weight: 1.0

[ZER-0-NE]

I am getting another error even after i add this line.

TypeError: separable_convolution2d() got an unexpected keyword argument 'data_format'

[Manugoyal123]

Try to reduce the height and width of image_resizer, in my case, I reduced them from 1200 to 600.