samples/quant_model/mobilenet_v1.c

// Mobilenet_v1_0.25_224 quant model
// MlModel struct initialization to include model I/O info.
// Bytecode loading, input/output processes.

#include <springbok.h>

#include "iree/base/api.h"
#include "iree/hal/api.h"
#include "samples/util/util.h"
#include "mobilenet_v1.h"

// Compiled module embedded here to avoid file IO:
#include "samples/quant_model/mobilenet_quant_input_c.h"
#include "samples/quant_model/mobilenet_v1_bytecode_module_dylib_c.h"

const MlModel kModel = {
    .num_input = 1,
    .num_input_dim = {4},
    .input_shape = {{1, 224, 224, 3}},
    .input_length = {224 * 224 * 3},
    .input_size_bytes = {sizeof(uint8_t)},
    .num_output = 1,
    .output_length = {1001},
    .output_size_bytes = sizeof(uint8_t),
    .hal_element_type = IREE_HAL_ELEMENT_TYPE_UINT_8,
    .entry_func = "module.main",
    .model_name = "mobilenet_v1_0.25_224_quant",
};

MobilenetV1Output score;

const iree_const_byte_span_t load_bytecode_module_data() {
  const struct iree_file_toc_t *module_file_toc =
      samples_quant_model_mobilenet_v1_bytecode_module_dylib_create();
  return iree_make_const_byte_span(module_file_toc->data,
                                   module_file_toc->size);
}

iree_status_t load_input_data(const MlModel *model, void **buffer,
                              iree_byte_span_t **byte_span) {
  byte_span[0] = malloc(sizeof(iree_byte_span_t));
  *byte_span[0] =
      iree_make_byte_span(mobilenet_quant_input,
                          model->input_size_bytes[0] * model->input_length[0]);
  return iree_ok_status();
}

iree_status_t process_output(const MlModel *model,
                  iree_hal_buffer_mapping_t *buffers,
                  MlOutput *output) {
  iree_status_t result = iree_ok_status();
  // find the label index with best prediction
  int best_out = 0;
  int best_idx = -1;
  for (int i = 0; i < model->output_length[0]; ++i) {
    uint8_t out = ((uint8_t *)buffers[0].contents.data)[i];
    if (out > best_out) {
      best_out = out;
      best_idx = i;
    }
  }
  score.best_out = best_out;
  score.best_idx = best_idx;

  LOG_INFO("Image prediction result is: id: %d", best_idx + 1);

  output->result = &score;
  output->len = sizeof(score);
  return result;
}