261 lines
8.7 KiB
C
261 lines
8.7 KiB
C
/**
|
|
******************************************************************************
|
|
* @file model.h
|
|
* @author AST Embedded Analytics Research Platform
|
|
* @date 2025-05-20T15:23:57+0800
|
|
* @brief AI Tool Automatic Code Generator for Embedded NN computing
|
|
******************************************************************************
|
|
* @attention
|
|
*
|
|
* Copyright (c) 2025 STMicroelectronics.
|
|
* All rights reserved.
|
|
*
|
|
* This software is licensed under terms that can be found in the LICENSE file
|
|
* in the root directory of this software component.
|
|
* If no LICENSE file comes with this software, it is provided AS-IS.
|
|
******************************************************************************
|
|
*/
|
|
#ifndef AI_MODEL_H
|
|
#define AI_MODEL_H
|
|
|
|
#include "model_config.h"
|
|
#include "ai_platform.h"
|
|
|
|
/******************************************************************************/
|
|
#define AI_MODEL_MODEL_NAME "model"
|
|
#define AI_MODEL_ORIGIN_MODEL_NAME "model"
|
|
|
|
/******************************************************************************/
|
|
#define AI_MODEL_ACTIVATIONS_ALIGNMENT (4)
|
|
#define AI_MODEL_INPUTS_IN_ACTIVATIONS (4)
|
|
#define AI_MODEL_OUTPUTS_IN_ACTIVATIONS (4)
|
|
|
|
/******************************************************************************/
|
|
#define AI_MODEL_IN_NUM (1)
|
|
|
|
AI_DEPRECATED
|
|
#define AI_MODEL_IN \
|
|
ai_model_inputs_get(AI_HANDLE_NULL, NULL)
|
|
|
|
#define AI_MODEL_IN_SIZE { \
|
|
AI_MODEL_IN_1_SIZE, \
|
|
}
|
|
#define AI_MODEL_IN_SIZE_BYTES { \
|
|
AI_MODEL_IN_1_SIZE_BYTES, \
|
|
}
|
|
#define AI_MODEL_IN_1_FORMAT (AI_BUFFER_FORMAT_FLOAT)
|
|
#define AI_MODEL_IN_1_CHANNEL (1024)
|
|
#define AI_MODEL_IN_1_SIZE (1024)
|
|
#define AI_MODEL_IN_1_SIZE_BYTES (4096)
|
|
|
|
/******************************************************************************/
|
|
#define AI_MODEL_OUT_NUM (1)
|
|
|
|
AI_DEPRECATED
|
|
#define AI_MODEL_OUT \
|
|
ai_model_outputs_get(AI_HANDLE_NULL, NULL)
|
|
|
|
#define AI_MODEL_OUT_SIZE { \
|
|
AI_MODEL_OUT_1_SIZE, \
|
|
}
|
|
#define AI_MODEL_OUT_SIZE_BYTES { \
|
|
AI_MODEL_OUT_1_SIZE_BYTES, \
|
|
}
|
|
#define AI_MODEL_OUT_1_FORMAT (AI_BUFFER_FORMAT_FLOAT)
|
|
#define AI_MODEL_OUT_1_CHANNEL (3)
|
|
#define AI_MODEL_OUT_1_SIZE (3)
|
|
#define AI_MODEL_OUT_1_SIZE_BYTES (12)
|
|
|
|
/******************************************************************************/
|
|
#define AI_MODEL_N_NODES (35)
|
|
|
|
|
|
AI_API_DECLARE_BEGIN
|
|
|
|
/*!
|
|
* @defgroup model
|
|
* @brief Public neural network APIs
|
|
* @details This is the header for the network public APIs declarations
|
|
* for interfacing a generated network model.
|
|
* @details The public neural network APIs hide the structure of the network
|
|
* and offer a set of interfaces to create, initialize, query, configure,
|
|
* run and destroy a network instance.
|
|
* To handle this, an opaque handler to the network context is provided
|
|
* on creation.
|
|
* The APIs are meant as stadard interfaces for the calling code; depending on
|
|
* the supported platforms and the models, different implementations could be
|
|
* available.
|
|
*/
|
|
|
|
/******************************************************************************/
|
|
/*! Public API Functions Declarations */
|
|
|
|
/*!
|
|
* @brief Get network library info as a datastruct.
|
|
* @ingroup model
|
|
* @param[in] network: the handler to the network context
|
|
* @param[out] report a pointer to the report struct where to
|
|
* store network info. See @ref ai_network_report struct for details
|
|
* @return a boolean reporting the exit status of the API
|
|
*/
|
|
AI_DEPRECATED
|
|
AI_API_ENTRY
|
|
ai_bool ai_model_get_info(
|
|
ai_handle network, ai_network_report* report);
|
|
|
|
|
|
|
|
/*!
|
|
* @brief Get network library report as a datastruct.
|
|
* @ingroup model
|
|
* @param[in] network: the handler to the network context
|
|
* @param[out] report a pointer to the report struct where to
|
|
* store network info. See @ref ai_network_report struct for details
|
|
* @return a boolean reporting the exit status of the API
|
|
*/
|
|
AI_API_ENTRY
|
|
ai_bool ai_model_get_report(
|
|
ai_handle network, ai_network_report* report);
|
|
|
|
|
|
/*!
|
|
* @brief Get first network error code.
|
|
* @ingroup model
|
|
* @details Get an error code related to the 1st error generated during
|
|
* network processing. The error code is structure containing an
|
|
* error type indicating the type of error with an associated error code
|
|
* Note: after this call the error code is internally reset to AI_ERROR_NONE
|
|
* @param network an opaque handle to the network context
|
|
* @return an error type/code pair indicating both the error type and code
|
|
* see @ref ai_error for struct definition
|
|
*/
|
|
AI_API_ENTRY
|
|
ai_error ai_model_get_error(ai_handle network);
|
|
|
|
|
|
/*!
|
|
* @brief Create a neural network.
|
|
* @ingroup model
|
|
* @details Instantiate a network and returns an object to handle it;
|
|
* @param network an opaque handle to the network context
|
|
* @param network_config a pointer to the network configuration info coded as a
|
|
* buffer
|
|
* @return an error code reporting the status of the API on exit
|
|
*/
|
|
AI_API_ENTRY
|
|
ai_error ai_model_create(
|
|
ai_handle* network, const ai_buffer* network_config);
|
|
|
|
|
|
/*!
|
|
* @brief Destroy a neural network and frees the allocated memory.
|
|
* @ingroup model
|
|
* @details Destroys the network and frees its memory. The network handle is returned;
|
|
* if the handle is not NULL, the unloading has not been successful.
|
|
* @param network an opaque handle to the network context
|
|
* @return an object handle : AI_HANDLE_NULL if network was destroyed
|
|
* correctly. The same input network handle if destroy failed.
|
|
*/
|
|
AI_API_ENTRY
|
|
ai_handle ai_model_destroy(ai_handle network);
|
|
|
|
|
|
/*!
|
|
* @brief Initialize the data structures of the network.
|
|
* @ingroup model
|
|
* @details This API initialized the network after a successfull
|
|
* @ref ai_model_create. Both the activations memory buffer
|
|
* and params (i.e. weights) need to be provided by caller application
|
|
*
|
|
* @param network an opaque handle to the network context
|
|
* @param params the parameters of the network (required).
|
|
* see @ref ai_network_params struct for details
|
|
* @return true if the network was correctly initialized, false otherwise
|
|
* in case of error the error type could be queried by
|
|
* using @ref ai_model_get_error
|
|
*/
|
|
AI_API_ENTRY
|
|
ai_bool ai_model_init(
|
|
ai_handle network, const ai_network_params* params);
|
|
|
|
|
|
/*!
|
|
* @brief Create and initialize a neural network (helper function)
|
|
* @ingroup model
|
|
* @details Helper function to instantiate and to initialize a network. It returns an object to handle it;
|
|
* @param network an opaque handle to the network context
|
|
* @param activations array of addresses of the activations buffers
|
|
* @param weights array of addresses of the weights buffers
|
|
* @return an error code reporting the status of the API on exit
|
|
*/
|
|
AI_API_ENTRY
|
|
ai_error ai_model_create_and_init(
|
|
ai_handle* network, const ai_handle activations[], const ai_handle weights[]);
|
|
|
|
|
|
/*!
|
|
* @brief Get network inputs array pointer as a ai_buffer array pointer.
|
|
* @ingroup model
|
|
* @param network an opaque handle to the network context
|
|
* @param n_buffer optional parameter to return the number of outputs
|
|
* @return a ai_buffer pointer to the inputs arrays
|
|
*/
|
|
AI_API_ENTRY
|
|
ai_buffer* ai_model_inputs_get(
|
|
ai_handle network, ai_u16 *n_buffer);
|
|
|
|
|
|
/*!
|
|
* @brief Get network outputs array pointer as a ai_buffer array pointer.
|
|
* @ingroup model
|
|
* @param network an opaque handle to the network context
|
|
* @param n_buffer optional parameter to return the number of outputs
|
|
* @return a ai_buffer pointer to the outputs arrays
|
|
*/
|
|
AI_API_ENTRY
|
|
ai_buffer* ai_model_outputs_get(
|
|
ai_handle network, ai_u16 *n_buffer);
|
|
|
|
|
|
/*!
|
|
* @brief Run the network and return the output
|
|
* @ingroup model
|
|
*
|
|
* @details Runs the network on the inputs and returns the corresponding output.
|
|
* The size of the input and output buffers is stored in this
|
|
* header generated by the code generation tool. See AI_MODEL_*
|
|
* defines into file @ref model.h for all network sizes defines
|
|
*
|
|
* @param network an opaque handle to the network context
|
|
* @param[in] input buffer with the input data
|
|
* @param[out] output buffer with the output data
|
|
* @return the number of input batches processed (default 1) or <= 0 if it fails
|
|
* in case of error the error type could be queried by
|
|
* using @ref ai_model_get_error
|
|
*/
|
|
AI_API_ENTRY
|
|
ai_i32 ai_model_run(
|
|
ai_handle network, const ai_buffer* input, ai_buffer* output);
|
|
|
|
|
|
/*!
|
|
* @brief Runs the network on the inputs.
|
|
* @ingroup model
|
|
*
|
|
* @details Differently from @ref ai_network_run, no output is returned, e.g. for
|
|
* temporal models with a fixed step size.
|
|
*
|
|
* @param network the network to be run
|
|
* @param[in] input buffer with the input data
|
|
* @return the number of input batches processed (usually 1) or <= 0 if it fails
|
|
* in case of error the error type could be queried by
|
|
* using @ref ai_model_get_error
|
|
*/
|
|
AI_API_ENTRY
|
|
ai_i32 ai_model_forward(
|
|
ai_handle network, const ai_buffer* input);
|
|
|
|
AI_API_DECLARE_END
|
|
|
|
#endif /* AI_MODEL_H */
|