diff --git a/include/aidge/operator/Resize.hpp b/include/aidge/operator/Resize.hpp
new file mode 100644
index 0000000000000000000000000000000000000000..04ce7731f95da8319870832b7c025c8d6c38e9d1
--- /dev/null
+++ b/include/aidge/operator/Resize.hpp
@@ -0,0 +1,87 @@
+/********************************************************************************
+ * Copyright (c) 2023 CEA-List
+ *
+ * This program and the accompanying materials are made available under the
+ * terms of the Eclipse Public License 2.0 which is available at
+ * http://www.eclipse.org/legal/epl-2.0.
+ *
+ * SPDX-License-Identifier: EPL-2.0
+ *
+ ********************************************************************************/
+
+#ifndef AIDGE_CORE_OPERATOR_Resize_H_
+#define AIDGE_CORE_OPERATOR_Resize_H_
+
+#include <cassert>
+#include <memory>
+#include <vector>
+
+#include "aidge/backend/OperatorImpl.hpp"
+#include "aidge/graph/Node.hpp"
+#include "aidge/operator/OperatorTensor.hpp"
+#include "aidge/utils/Registrar.hpp"
+#include "aidge/utils/StaticAttributes.hpp"
+#include "aidge/utils/Types.h"
+
+namespace Aidge {
+
+class Resize_Op : public OperatorTensor,
+ public Registrable<Resize_Op, std::string, std::unique_ptr<OperatorImpl>(const Resize_Op&)>
+ {
+
+public:
+ static const std::string Type; // du
+
+ Resize_Op() : OperatorTensor(Type, 2, 0, 1) {}
+ // 4 input 0 attribute 1 output
+ // "data_input", "roi", "scales", "sizes"
+
+ /**
+ * @brief Copy-constructor. Copy the operator attributes and its output tensor(s), but not its input tensors (the new operator has no input associated).
+ * @param op Operator to copy.
+ */
+ Resize_Op(const Resize_Op& op)
+ : OperatorTensor(op)
+ {
+ // copy an operator
+ if (!op.backend().empty()) {
+ SET_IMPL_MACRO(Resize_Op, *this, op.backend());
+ }
+ else {
+ mImpl = nullptr;
+ }
+ }
+
+ /**
+ * @brief Clone the operator using its copy-constructor.
+ * @see Operator::Resize_Op
+ */
+ std::shared_ptr<Operator> clone() const override {
+ return std::make_shared<Resize_Op>(*this);
+ }
+
+ // function see inputs
+ bool forwardDims(bool allowDataDependency = false) override final;
+
+ void setBackend(const std::string& name, DeviceIdx_t device = 0) override final{
+ // // to select which backend CUDA / CPU?
+ // mImpl = Registrar<Resize_Op>::create(name)(*this);
+ // mOutputs[0]->setBackend(name, device);
+ }
+
+ static const std::vector<std::string> getInputsName(){
+ return {"data_input", "roi ", "scales", "sizes"};
+ }
+ static const std::vector<std::string> getOutputsName(){
+ return {"data_output"};
+ }
+};
+
+// WTF
+inline std::shared_ptr<Node> Resize(const std::string &name = "") {
+ // FIXME: properly handle default w&b initialization in every cases
+ return std::make_shared<Node>(std::make_shared<Resize_Op>(), name);
+}
+} // namespace Aidge
+
+#endif /* AIDGE_CORE_OPERATOR_Resize_H_ */
diff --git a/python_binding/operator/pybind_Resize.cpp b/python_binding/operator/pybind_Resize.cpp
new file mode 100644
index 0000000000000000000000000000000000000000..cdaefebb866a7c440a6a6dc0a8d33bce818b5f20
--- /dev/null
+++ b/python_binding/operator/pybind_Resize.cpp
@@ -0,0 +1,27 @@
+/********************************************************************************
+ * Copyright (c) 2023 CEA-List
+ *
+ * This program and the accompanying materials are made available under the
+ * terms of the Eclipse Public License 2.0 which is available at
+ * http://www.eclipse.org/legal/epl-2.0.
+ *
+ * SPDX-License-Identifier: EPL-2.0
+ *
+ ********************************************************************************/
+
+#include <pybind11/pybind11.h>
+
+#include "aidge/operator/Resize.hpp"
+#include "aidge/operator/OperatorTensor.hpp"
+
+namespace py = pybind11;
+namespace Aidge {
+
+void init_Resize(py::module& m) {
+ py::class_<Resize_Op, std::shared_ptr<Resize_Op>, OperatorTensor>(m, "ResizeOp", py::multiple_inheritance())
+ .def("get_inputs_name", &Resize_Op::getInputsName)
+ .def("get_outputs_name", &Resize_Op::getOutputsName);
+
+ m.def("Resize", &Resize, py::arg("name") = "");
+}
+} // namespace Aidge
diff --git a/src/operator/Resize.cpp b/src/operator/Resize.cpp
new file mode 100644
index 0000000000000000000000000000000000000000..32c8a88939e1d0cfb180fc7fe083bcfe81ede0d1
--- /dev/null
+++ b/src/operator/Resize.cpp
@@ -0,0 +1,105 @@
+/********************************************************************************
+ * Copyright (c) 2023 CEA-List
+ *
+ * This program and the accompanying materials are made available under the
+ * terms of the Eclipse Public License 2.0 which is available at
+ * http://www.eclipse.org/legal/epl-2.0.
+ *
+ * SPDX-License-Identifier: EPL-2.0
+ *
+ ********************************************************************************/
+
+#include "aidge/operator/Resize.hpp"
+
+#include <cstddef> // std::size_t
+#include <cstdint> // std::int64_t
+#include <stdexcept> // std::runtime_error
+#include <string>
+#include <vector>
+#include <fmt/core.h>
+
+#include "aidge/backend/OperatorImpl.hpp"
+#include "aidge/data/Tensor.hpp"
+#include "aidge/utils/ErrorHandling.hpp"
+#include "aidge/utils/Types.h"
+
+
+const std::string Aidge::Resize_Op::Type = "Resize";
+
+bool Aidge::Resize_Op::forwardDims(bool /*allowDataDependency*/) {
+
+ // check input ("data_input","roi", "scales", "data_input","sizes") has been associated
+ for (size_t i = 0; i < 4; ++i) {
+ if (!getInput(i)) {
+ AIDGE_THROW_OR_ABORT(std::runtime_error, "{}: input #{} not provided", type(), i);
+ }
+ }
+
+ // roi not implemented
+
+ // require scales of the input to have dims defined
+
+ if ((!getInput(0)->empty()) && (getInput(1)->empty() && !getInput(2)->empty() && getInput(3)->empty()) ) {
+ // condition 1. input data_input tensor + scales
+
+
+ std::vector<DimSize_t> outDims = getInput(0)->dims();
+ fmt::print("input dims 0 = {}\n", outDims);
+ fmt::print("input dims 1 = {}\n", getInput(1)->dims());
+ fmt::print("input dims 2 = {}\n", getInput(2)->dims());
+ fmt::print("input dims 3 = {}\n", getInput(3)->dims());
+
+ printf("Scales Input(1)");
+
+ float* scales = static_cast<float*>(getInput(1)->getImpl()->rawPtr()); //TODO cast according to tensor type
+
+ // std::vector<DimSize_t> outDims[ width_out = width_input_tensor*scales[3], \
+ // height_out = heigh_input_tensor*scales[2], \
+ // depth_input_tensor = depth_input_tensor*scales[1] \
+ // batch_input_tensor = batch_input_tensor*scales[0] ];
+
+ for (std::size_t dim=0; dim < getInput(1)->size(); ++dim) {
+ // todo verify if batch and depth is not 1 !
+ printf("{} {} \n", dim, scales[dim] );
+ outDims[dim] *= scales[dim];
+ }
+
+ mOutputs[0]->resize(outDims);
+ fmt::print("output dims 0 = {}\n", outDims);
+
+ return true;
+ }
+
+ if ((!getInput(0)->empty()) && (getInput(1)->empty()) && !getInput(2)->empty()) {
+ // condition 2. input tensor + sizes
+ // to verify 2 arg
+ // "Input 0 and 2 must be provided and input 1 must not be provided."
+ // "data_input" and "sizes"
+
+ std::vector<DimSize_t> outDims = getInput(0)->dims();
+ fmt::print("input dims 0 = {}\n", outDims);
+ fmt::print("input dims 1 = {}\n", getInput(1)->dims());
+ fmt::print("input dims 2 = {}\n", getInput(2)->dims());
+ fmt::print("input dims 3 = {}\n", getInput(3)->dims());
+
+ printf("Sizes Input(2)");
+
+ float* sizes = static_cast<float*>(getInput(2)->getImpl()->rawPtr()); //TODO cast according to tensor type
+
+ // std::vector<DimSize_t> outDims[ width_out = sizes[0], \
+ // height_out = sizes[1], \
+ // depth_input_tensor, \
+ // batch_input_tensor];
+
+ for (std::size_t dim=0; dim < getInput(1)->size(); ++dim) {
+ // todo verify if batch and depth is not 1 !
+ outDims[dim] = sizes[dim];
+ printf("{} {} \n", dim, sizes[dim] );
+ }
+ mOutputs[0]->resize(outDims);
+ fmt::print("output dims 0 = {}\n", outDims);
+ return true;
+ }
+
+ return false;
+}
\ No newline at end of file