[MO] Remove Apache MxNet models support in openvino-dev

docs/articles_en/documentation/legacy-features.rst

@@ -78,7 +78,7 @@ offering.
 Discontinued:
 #############
 
-.. dropdown:: Apache MXNet, Caffe, and Kaldi model formats
+.. dropdown:: Caffe, and Kaldi model formats
 
    |   *New solution:* conversion to ONNX via external tools
    |   *Old solution:* model support discontinued with OpenVINO 2024.0

docs/articles_en/documentation/legacy-features/transition-legacy-conversion-api/legacy-conversion-api/[legacy]-model-optimizer-faq.rst

@@ -234,7 +234,7 @@ Q14. What does the message "Cannot infer shape for node {} because there is no C
 Q15. What does the message "Framework name can not be deduced from the given options. Use --framework to choose one of Caffe, TensorFlow, MXNet" mean?
 ######################################################################################################################################################
 
-**A:** You have run Model Optimizer without a flag ``--framework caffe|tf|mxnet``. Model Optimizer tries to deduce the framework by the extension of input model file (``.pb`` for TensorFlow, ``.caffemodel`` for Caffe, ``.params`` for Apache MXNet). Your input model might have a different extension and you need to explicitly set the source framework. For example, use ``--framework caffe``.
+**A:** You have run Model Optimizer without a flag ``--framework caffe|tf``. Model Optimizer tries to deduce the framework by the extension of input model file (``.pb`` for TensorFlow, ``.caffemodel`` for Caffe, ``.params`` for Apache MXNet). Your input model might have a different extension and you need to explicitly set the source framework. For example, use ``--framework caffe``.
 
 .. _question-16:
 

docs/dev/ov_dependencies.txt

@@ -147,7 +147,6 @@ mccabe
 mistune
 mongo-python-driver
 more-itertools
-mxnet
 mypy
 mypy-extensions
 networkx

docs/dev/pypi_publish/pypi-openvino-dev.md

@@ -83,7 +83,6 @@ pip install openvino-dev[extras]
 | :-------------------------------| :------------------------------------------------------------------------------- |
 | caffe                           |   [Caffe*](https://caffe.berkeleyvision.org/)                                    |
 | kaldi                           |   [Kaldi*](https://github.com/kaldi-asr/kaldi)                                   |
-| mxnet                           |   [Apache MXNet*](https://mxnet.apache.org/)                                     |
 | onnx                            |   [ONNX*](https://github.com/microsoft/onnxruntime/)                             |
 | pytorch                         |   [PyTorch*](https://pytorch.org/)                                               |
 | tensorflow                      |   [TensorFlow* 1.x](https://www.tensorflow.org/versions#tensorflow_1)            |
@@ -149,14 +148,14 @@ Users in China might encounter errors while downloading sources via PIP during O
 If you use zsh (Z shell) interpreter, that is the default shell for macOS starting with version 10.15 (Catalina), you may encounter the following error while installing `openvino-dev` package with extras:
 
 ```sh
-pip install openvino-dev[tensorflow2,mxnet,caffe]
-zsh: no matches found: openvino-dev[tensorflow2,mxnet,caffe]
+pip install openvino-dev[tensorflow2,caffe]
+zsh: no matches found: openvino-dev[tensorflow2,caffe]
 ```
 
 By default zsh interprets square brackets as an expression for pattern matching. To resolve this issue, you need to escape the command with quotes:
 
 ```sh
-pip install 'openvino-dev[tensorflow2,mxnet,caffe]'
+pip install 'openvino-dev[tensorflow2,caffe]'
 ```
 
 To avoid such issues you can also disable globbing for PIP commands by defining an alias in `~/.zshrc` file:

docs/sphinx_setup/_static/download/supported_models.csv

@@ -211,11 +211,6 @@ facenet,Object Detection,onnx,FP16,,,+
 facenet,Object Detection,onnx,FP16-INT8,,,+
 facenet-20180408-102900,Object Detection,tf,FP16,+,+,+
 facenet-20180408-102900,Object Detection,tf,FP32,+,+,
-face-recognition-mobilefacenet-arcface,Object Detection,mxnet,FP16,,,+
-face-recognition-mobilefacenet-arcface,Object Detection,mxnet,FP16-INT8,,,+
-face-recognition-resnet50-arcface,Object Detection,mxnet,FP16,+,+,+
-face-recognition-resnet50-aws,Object Detection,mxnet,FP16,,,+
-face-recognition-resnet50-aws,Object Detection,mxnet,FP32,,,+
 face-reidentification-retail-0095,Object Detection,onnx,FP16,,,+
 face-reidentification-retail-0095,Object Detection,onnx,FP16-INT8,,,+
 facial-landmarks-35-adas-0002,Object Detection,caffe,FP16,,,+
@@ -428,8 +423,6 @@ ocrnet-hrnet-w18,Text Detection,paddle,FP16-INT8,+,+,
 ocrnet-hrnet-w18,Text Detection,paddle,FP32,+,,
 ocrnet-hrnet-w48,Text Detection,paddle,FP16,+,,
 ocrnet-hrnet-w48,Text Detection,paddle,FP32,+,,
-octave-resnext-101-0.25,Image Classification,mxnet,FP16,+,+,+
-octave-resnext-101-0.25,Image Classification,mxnet,FP32,+,,
 openchat-3.6-8b-20240522,Large Language Model,pytorch,intel-optimum default,,+,
 open-closed-eye-0001,Image Classification,onnx,FP16,,,+
 open-closed-eye-0001,Image Classification,onnx,FP16-INT8,,,+
@@ -629,8 +622,6 @@ squeezenet1.0-12,Image Classification,onnx,FP16,,,+
 squeezenet1.0-12,Image Classification,onnx,FP32,+,+,+
 squeezenet1.1-caffe,Image Classification,caffe,FP16,+,+,+
 squeezenet1.1-caffe,Image Classification,caffe,FP32,+,+,+
-squeezenet1.1-mxnet,Image Classification,mxnet,FP16,+,+,+
-squeezenet1.1-mxnet,Image Classification,mxnet,FP32,+,+,+
 squeezenet1.1-onnx,Image Classification,onnx,FP32,+,,
 srgan-onnx,"Image Processing, Enhancement",onnx,FP16,+,,+
 srgan-tf,"Image Processing, Enhancement",tf,FP16,+,+,

src/tests/test_utils/functional_test_utils/layer_tests_summary/conformance_helper_tools/conformance_vs_accuracy_comparator.py

@@ -38,7 +38,7 @@ def get_precision(self):
         return self.__model_prc
 
 def path_to_model(model_path: os.path, prefix: str):
-    frameworks = {'tf', 'tf2', 'caffe', 'onnx', 'mxnet', 'paddle', 'kaldi'}
+    frameworks = {'tf', 'tf2', 'caffe', 'onnx', 'paddle', 'kaldi'}
     precisions = {'FP16', 'FP32', 'INT8', 'INT1'}
     # remove share path + model.xml
     model_path = model_path.replace('\n', '')

tools/constraints.txt

@@ -3,8 +3,6 @@
 # files because the version differs between them:
 # tensorflow, numpy
 h5py>=3.1.0,<3.11.0
-mxnet~=1.2.0; sys_platform == 'win32'
-mxnet>=1.7.0.post2,<=1.9.1; sys_platform != 'win32'
 onnx>=1.8.1,<=1.16.0
 networkx<=3.1.0
 pytest>=5.0,<8.4

tools/mo/README.md

@@ -18,10 +18,6 @@ E.g. the command below will install dependencies to support ONNX\* and TensorFlo
 ```
 pip install openvino-dev[onnx,tensorflow2]
 ```
-To enable support of MxNet\* models run:
-```
-pip install openvino-dev[mxnet]
-```
 To enable support of all frameworks:
 ```
 pip install openvino-dev[all]

tools/mo/automation/package_BOM.txt

@@ -252,111 +252,6 @@ openvino/tools/mo/front/LayerNorm.py
 openvino/tools/mo/front/Log1p.py
 openvino/tools/mo/front/MatMul_normalizer.py
 openvino/tools/mo/front/MoveEmbeddedInputsToInputs.py
-openvino/tools/mo/front/mxnet/__init__.py
-openvino/tools/mo/front/mxnet/activation.py
-openvino/tools/mo/front/mxnet/adaptive_avg_pooling_ext.py
-openvino/tools/mo/front/mxnet/add_input_data_to_prior_boxes.py
-openvino/tools/mo/front/mxnet/arange_ext.py
-openvino/tools/mo/front/mxnet/arange_like_ext.py
-openvino/tools/mo/front/mxnet/arange_like_replacer.py
-openvino/tools/mo/front/mxnet/arange_replacer.py
-openvino/tools/mo/front/mxnet/batch_dot_ext.py
-openvino/tools/mo/front/mxnet/block_grad_ext.py
-openvino/tools/mo/front/mxnet/box_nms_ext.py
-openvino/tools/mo/front/mxnet/cast_ext.py
-openvino/tools/mo/front/mxnet/check_softmax_node_inputs.py
-openvino/tools/mo/front/mxnet/clip_ext.py
-openvino/tools/mo/front/mxnet/conv_ext.py
-openvino/tools/mo/front/mxnet/copy_ext.py
-openvino/tools/mo/front/mxnet/crop_ext.py
-openvino/tools/mo/front/mxnet/cumsum.py
-openvino/tools/mo/front/mxnet/cumsum_ext.py
-openvino/tools/mo/front/mxnet/custom.py
-openvino/tools/mo/front/mxnet/custom_rpn_proposal.py
-openvino/tools/mo/front/mxnet/deformable_conv_ext.py
-openvino/tools/mo/front/mxnet/deformable_psroi_pooling_ext.py
-openvino/tools/mo/front/mxnet/div_sqrt_dim.py
-openvino/tools/mo/front/mxnet/dropout_ext.py
-openvino/tools/mo/front/mxnet/einsum_ext.py
-openvino/tools/mo/front/mxnet/elementwise_ext.py
-openvino/tools/mo/front/mxnet/eltwise_scalar_replacers.py
-openvino/tools/mo/front/mxnet/exp_ext.py
-openvino/tools/mo/front/mxnet/expand_dims_ext.py
-openvino/tools/mo/front/mxnet/extractor.py
-openvino/tools/mo/front/mxnet/extractors/__init__.py
-openvino/tools/mo/front/mxnet/extractors/add_n.py
-openvino/tools/mo/front/mxnet/extractors/batchnorm.py
-openvino/tools/mo/front/mxnet/extractors/concat.py
-openvino/tools/mo/front/mxnet/extractors/l2_normalization.py
-openvino/tools/mo/front/mxnet/extractors/multibox_prior.py
-openvino/tools/mo/front/mxnet/extractors/relu.py
-openvino/tools/mo/front/mxnet/extractors/scaleshift.py
-openvino/tools/mo/front/mxnet/extractors/slice_axis.py
-openvino/tools/mo/front/mxnet/extractors/utils.py
-openvino/tools/mo/front/mxnet/eye_ext.py
-openvino/tools/mo/front/mxnet/eye_mx_to_eye.py
-openvino/tools/mo/front/mxnet/fft_ext.py
-openvino/tools/mo/front/mxnet/flatten_ext.py
-openvino/tools/mo/front/mxnet/fully_connected_ext.py
-openvino/tools/mo/front/mxnet/gather.py
-openvino/tools/mo/front/mxnet/gather_ext.py
-openvino/tools/mo/front/mxnet/gluoncv_ssd_anchors.py
-openvino/tools/mo/front/mxnet/instance_norm_ext.py
-openvino/tools/mo/front/mxnet/layer_norm_ext.py
-openvino/tools/mo/front/mxnet/leaky_relu.py
-openvino/tools/mo/front/mxnet/loader.py
-openvino/tools/mo/front/mxnet/lrn_ext.py
-openvino/tools/mo/front/mxnet/max_ext.py
-openvino/tools/mo/front/mxnet/modulated_deformable_conv_ext.py
-openvino/tools/mo/front/mxnet/modulated_deformable_conv_replacer.py
-openvino/tools/mo/front/mxnet/multibox_detection_ext.py
-openvino/tools/mo/front/mxnet/mx_reshape_reverse.py
-openvino/tools/mo/front/mxnet/mx_reshape_to_reshape.py
-openvino/tools/mo/front/mxnet/MXFFTToDFT.py
-openvino/tools/mo/front/mxnet/MXRepeatReplacer.py
-openvino/tools/mo/front/mxnet/nd_to_params.py
-openvino/tools/mo/front/mxnet/null_ext.py
-openvino/tools/mo/front/mxnet/pad_ext.py
-openvino/tools/mo/front/mxnet/pooling_ext.py
-openvino/tools/mo/front/mxnet/proposal_ext.py
-openvino/tools/mo/front/mxnet/psroi_pooling_ext.py
-openvino/tools/mo/front/mxnet/random_uniform_ext.py
-openvino/tools/mo/front/mxnet/register_custom_ops.py
-openvino/tools/mo/front/mxnet/repeat_ext.py
-openvino/tools/mo/front/mxnet/reshape_ext.py
-openvino/tools/mo/front/mxnet/RNN_ext.py
-openvino/tools/mo/front/mxnet/rnn_param_concat.py
-openvino/tools/mo/front/mxnet/roi_pooling_ext.py
-openvino/tools/mo/front/mxnet/roll_ext.py
-openvino/tools/mo/front/mxnet/shape_array_ext.py
-openvino/tools/mo/front/mxnet/sigmoid.py
-openvino/tools/mo/front/mxnet/slice_channel_ext.py
-openvino/tools/mo/front/mxnet/slice_ext.py
-openvino/tools/mo/front/mxnet/slice_like_ext.py
-openvino/tools/mo/front/mxnet/slice_replacers.py
-openvino/tools/mo/front/mxnet/softmax.py
-openvino/tools/mo/front/mxnet/softmax_activation_ext.py
-openvino/tools/mo/front/mxnet/softmax_ext.py
-openvino/tools/mo/front/mxnet/softmax_output_ext.py
-openvino/tools/mo/front/mxnet/softsign_ext.py
-openvino/tools/mo/front/mxnet/squeeze_ext.py
-openvino/tools/mo/front/mxnet/ssd_anchor_reshape.py
-openvino/tools/mo/front/mxnet/ssd_detection_output_replacer.py
-openvino/tools/mo/front/mxnet/ssd_pattern_flatten_softmax_activation.py
-openvino/tools/mo/front/mxnet/ssd_pattern_remove_flatten.py
-openvino/tools/mo/front/mxnet/ssd_pattern_remove_reshape.py
-openvino/tools/mo/front/mxnet/ssd_pattern_remove_transpose.py
-openvino/tools/mo/front/mxnet/ssd_reorder_detection_out_inputs.py
-openvino/tools/mo/front/mxnet/stack_ext.py
-openvino/tools/mo/front/mxnet/swapaxis_ext.py
-openvino/tools/mo/front/mxnet/take_ext.py
-openvino/tools/mo/front/mxnet/tile_ext.py
-openvino/tools/mo/front/mxnet/tile_replacer.py
-openvino/tools/mo/front/mxnet/transpose_ext.py
-openvino/tools/mo/front/mxnet/up_sampling_ext.py
-openvino/tools/mo/front/mxnet/where_ext.py
-openvino/tools/mo/front/mxnet/yolo_v3_mobilenet1_voc.json
-openvino/tools/mo/front/mxnet/zeros_ext.py
 openvino/tools/mo/front/no_op_eraser.py
 openvino/tools/mo/front/non_max_suppression_normalize.py
 openvino/tools/mo/front/OneHotDepthNormalizer.py
@@ -692,16 +587,13 @@ openvino/tools/mo/load/caffe/loader.py
 openvino/tools/mo/load/kaldi/__init__.py
 openvino/tools/mo/load/kaldi/loader.py
 openvino/tools/mo/load/loader.py
-openvino/tools/mo/load/mxnet/__init__.py
-openvino/tools/mo/load/mxnet/loader.py
 openvino/tools/mo/load/onnx/__init__.py
 openvino/tools/mo/load/onnx/loader.py
 openvino/tools/mo/load/tf/__init__.py
 openvino/tools/mo/load/tf/loader.py
 openvino/tools/mo/main.py
 openvino/tools/mo/main_caffe.py
 openvino/tools/mo/main_kaldi.py
-openvino/tools/mo/main_mxnet.py
 openvino/tools/mo/main_onnx.py
 openvino/tools/mo/main_paddle.py
 openvino/tools/mo/main_tf.py
@@ -827,7 +719,6 @@ openvino/tools/mo/middle/UselessSplitEraser.py
 openvino/tools/mo/mo.py
 openvino/tools/mo/mo_caffe.py
 openvino/tools/mo/mo_kaldi.py
-openvino/tools/mo/mo_mxnet.py
 openvino/tools/mo/mo_onnx.py
 openvino/tools/mo/mo_paddle.py
 openvino/tools/mo/mo_tf.py
@@ -845,7 +736,6 @@ openvino/tools/mo/ops/__init__.py
 openvino/tools/mo/ops/activation.py
 openvino/tools/mo/ops/activation_ops.py
 openvino/tools/mo/ops/adaptive_avg_pooling.py
-openvino/tools/mo/ops/arange_like.py
 openvino/tools/mo/ops/argmax.py
 openvino/tools/mo/ops/argmin.py
 openvino/tools/mo/ops/assert_op.py
@@ -882,7 +772,6 @@ openvino/tools/mo/ops/dequantize_linear.py
 openvino/tools/mo/ops/detection_output_onnx.py
 openvino/tools/mo/ops/DetectionOutput.py
 openvino/tools/mo/ops/dft.py
-openvino/tools/mo/ops/div_sqrt_dim.py
 openvino/tools/mo/ops/dropoutmask.py
 openvino/tools/mo/ops/einsum.py
 openvino/tools/mo/ops/elementwise.py
@@ -932,7 +821,6 @@ openvino/tools/mo/ops/multinomial.py
 openvino/tools/mo/ops/mvn.py
 openvino/tools/mo/ops/mxfft.py
 openvino/tools/mo/ops/mxrepeat.py
-openvino/tools/mo/ops/mxreshape.py
 openvino/tools/mo/ops/NextIteration.py
 openvino/tools/mo/ops/nms_rotated.py
 openvino/tools/mo/ops/non_max_suppression.py

tools/mo/openvino/tools/mo/analysis/inputs.py

@@ -37,21 +37,6 @@ def fifo_queue_analysis(cls, graph: Graph, inputs_desc: dict):
                                                                              'data_type': fifo_queue.types[port_ind]}
         return inputs_to_ignore
 
-    @classmethod
-    def ignore_mxnet_softmax_inputs(cls, graph: Graph):
-        """
-        MxNet Softmax layers may have additional inputs which should be ignored. Refer to the
-        openvino/tools/mo/front/mxnet/check_softmax_node_inputs.py.
-        """
-        inputs_to_ignore = set()
-        softmax_nodes = []
-        [softmax_nodes.extend(graph.get_op_nodes(op=op)) for op in ('SoftMax', 'SoftmaxActivation', 'SoftmaxOutput')]
-        for softmax_node in softmax_nodes:
-            for i in range(1, len(softmax_node.in_nodes())):
-                if softmax_node.in_node(i).has_valid('op') and softmax_node.in_node(i).op == 'Parameter':
-                    inputs_to_ignore.add(softmax_node.in_node(i).id)
-        return inputs_to_ignore
-
     @classmethod
     def iterator_get_next_analysis(cls, graph: Graph, inputs_desc: dict):
         message = None
@@ -80,9 +65,6 @@ def analyze(self, graph: Graph):
         inputs_desc = dict()
         message = InputsAnalysis.iterator_get_next_analysis(graph, inputs_desc)
         inputs_to_ignore = InputsAnalysis.fifo_queue_analysis(graph, inputs_desc)
-        if graph.graph['fw'] == 'mxnet':
-            inputs_to_ignore.update(InputsAnalysis.ignore_mxnet_softmax_inputs(graph))
-
         inputs = graph.get_op_nodes(op='Parameter')
         for input in inputs:
             inputs_desc[input.name] = {'shape': input.soft_get('shape', None),

tools/mo/openvino/tools/mo/convert.py

@@ -59,14 +59,6 @@ def convert_model(
         tensorboard_logdir: [str, pathlib.Path] = None,
         tensorflow_custom_layer_libraries: [str, pathlib.Path] = None,
 
-        # MXNet-specific parameters:
-        input_symbol: [str, pathlib.Path] = None,
-        nd_prefix_name: str = None,
-        pretrained_model_name: str = None,
-        save_params_from_nd: bool = None,
-        legacy_mxnet_model: bool = None,
-        enable_ssd_gluoncv: bool = False,
-
         # Caffe*-specific parameters:
         input_proto: [str, pathlib.Path] = None,
         caffe_parser_path: [str, pathlib.Path] = None,
@@ -129,7 +121,7 @@ def convert_model(
             a string or list of strings of the following format. Quoted list of comma-separated
             input nodes names with shapes, data types, and values for freezing.
             If operation names are specified, the order of inputs in converted
-            model will be the same as order of specified operation names (applicable for TF2, ONNX, MxNet).
+            model will be the same as order of specified operation names (applicable for TF2, ONNX).
             The shape and value are specified as comma-separated lists. The data type of input node is specified
             in braces and can have one of the values: f64 (float64), f32 (float32), f16 (float16), i64
             (int64), i32 (int32), u8 (uint8), boolean (bool). Data type is optional.
@@ -308,24 +300,6 @@ def convert_model(
             TensorFlow*: comma separated list of shared libraries with TensorFlow*
             custom operations implementation.
 
-    MXNet-specific parameters:
-        :param input_symbol:
-            Symbol file (for example, model-symbol.json) that contains a topology
-            structure and layer attributes
-        :param nd_prefix_name:
-            Prefix name for args.nd and argx.nd files.
-        :param pretrained_model_name:
-            Name of a pretrained MXNet model without extension and epoch number.
-            This model will be merged with args.nd and argx.nd files
-        :param save_params_from_nd:
-            Enable saving built parameters file from .nd files
-        :param legacy_mxnet_model:
-            Enable MXNet loader to make a model compatible with the latest MXNet
-            version. Use only if your model was trained with MXNet version lower
-            than 1.0.0
-        :param enable_ssd_gluoncv:
-            Enable pattern matchers replacers for converting gluoncv ssd topologies.
-
     Caffe*-specific parameters:
         :param input_proto:
             Deploy-ready prototxt file that contains a topology structure and