coco_detection_metrics and pascal_voc_detection_metrics mismatch

System information

• Have I written custom code (as opposed to using a stock example script provided in TensorFlow):
• OS Platform and Distribution (e.g., Linux Ubuntu 16.04):
• Mobile device (e.g., Pixel 4, Samsung Galaxy 10) if the issue happens on mobile device:
• TensorFlow installed from (source or binary):
• TensorFlow version (use command below):
• Python version:
• Bazel version (if compiling from source):
• GCC/Compiler version (if compiling from source):
• CUDA/cuDNN version:
• GPU model and memory:

Please provide the entire URL of the model you are using?

https://github.com/tensorflow/models/tree/master/research/object_detection
Describe the current behavior
When I run eval with coco_detection_metrics I have AP (IOU=0.5)=0.41
When I run eval with pascal_voc_detection_metrics I have AP (IOU=0.5) = 0.86
Describe the expected behavior
This metrics must be equal
Code to reproduce the issue

config:
model {
faster_rcnn {
num_classes: 2
image_resizer {
keep_aspect_ratio_resizer {
min_dimension: 400
max_dimension: 400
}
}
feature_extractor {
type: 'faster_rcnn_resnet50'
first_stage_features_stride: 8
}
first_stage_anchor_generator {
grid_anchor_generator {
scales: [0.25, 0.5]
aspect_ratios: [0.5, 1.0]
height_stride: 8
width_stride: 8
}
}
first_stage_box_predictor_conv_hyperparams {
op: CONV
regularizer {
l2_regularizer {
weight: 0.0
}
}
initializer {
truncated_normal_initializer {
stddev: 0.01
}
}
}
first_stage_nms_score_threshold: 0.0
first_stage_nms_iou_threshold: 0.0
first_stage_localization_loss_weight: 2.0
first_stage_objectness_loss_weight: 1.0
initial_crop_size: 14
maxpool_kernel_size: 2
maxpool_stride: 2
second_stage_box_predictor {
mask_rcnn_box_predictor {
use_dropout: false
dropout_keep_probability: 1.0
fc_hyperparams {
op: FC
regularizer {
l2_regularizer {
weight: 0.0
}
}
initializer {
variance_scaling_initializer {
factor: 1.0
uniform: true
mode: FAN_AVG
}
}
}
}
}
second_stage_post_processing {
batch_non_max_suppression {
score_threshold: 0.0
iou_threshold: 0.0
}
score_converter: SOFTMAX
}
second_stage_localization_loss_weight: 2.0
second_stage_classification_loss_weight: 1.0
}
}

train_config: {
batch_size: 4
optimizer {
momentum_optimizer: {
learning_rate: {
manual_step_learning_rate {
initial_learning_rate: 0.0003
schedule {
step: 40000
learning_rate: .00003
}
schedule {
step: 510000
learning_rate: .000003
}
}
}
momentum_optimizer_value: 0.9
}
use_moving_average: false
}
gradient_clipping_by_norm: 10.0
fine_tune_checkpoint: ""
from_detection_checkpoint: true

num_steps: 2000
data_augmentation_options {
}

train_input_reader: {
tf_record_input_reader {
input_path: ""
}
label_map_path: ""

}

eval_config: {
num_examples: 30
max_evals: 1
num_visualizations: 30
eval_interval_secs: 900
use_moving_averages: false
metrics_set: "coco_detection_metrics"
}

eval_input_reader: {
tf_record_input_reader {
input_path: ""
}
label_map_path: ""
shuffle: false
num_readers: 1
}
Other info / logs