[VLM] Image resize model

[yatarkan]

No description provided.

[ilya-lavrenov]

I have similar code for image to image / inpainting scenario https://github.com/ilya-lavrenov/openvino.genai/blob/ce3e1e3b095c2dc3d0e613c15bfa5fe778285d77/src/cpp/src/image_generation/image_processor.hpp#L105, does my implementation provide the same results? I'm not sure why some clamp, round operations are needed, my implementation is taken from translator of pytorch operation to OpenVINO in PT FE.

[ilya-lavrenov]

Now, it's a part of master

openvino.genai/src/cpp/src/image_generation/image_processor.hpp

Lines 35 to 47 in aef1591

	class ImageResizer {
	public:
	ImageResizer(const std::string& device, ov::element::Type type, ov::Layout layout, ov::op::v11::Interpolate::InterpolateMode interpolation_mode);
	ov::Tensor execute(ov::Tensor image, int64_t dst_height, int64_t dst_width);
	private:
	size_t get_and_check_width_idx(const Layout& layout, const PartialShape& shape);
	size_t get_and_check_height_idx(const Layout& layout, const PartialShape& shape);
	ov::InferRequest m_request;
	};

CC @Wovchena can it be tried in minicpm or other VLMs?