Jetson Inference
DNN Vision Library

Utilities for automatically downloading pre-trained models. More...

Macros

#define JSON_STR(x)   (x.is_string() ? x.get<std::string>() : "")
 Helper macro for getting a string from a JSON element. More...
 
#define JSON_STR_DEFAULT(x, default)   (x.is_string() ? x.get<std::string>() : default)
 Helper macro for getting a string from a JSON element. More...
 

Functions

bool DownloadModel (nlohmann::json &model, uint32_t retries=2)
 Download a pre-trained model given its JSON config. More...
 
bool DownloadModel (const char *type, const char *name, nlohmann::json &model, uint32_t retries=2)
 Download a pre-trained model and return its JSON config. More...
 
bool FindModel (const char *type, const char *name)
 Check if a model can be found in the model manifest. More...
 
bool FindModel (const char *type, const char *name, nlohmann::json &model)
 Find a pre-trained model's config in the model manifest. More...
 
bool FindModel (const char *type, const char *name, nlohmann::json &models, nlohmann::json &model)
 Find a pre-trained model's config in the model manifest. More...
 
bool LoadModelManifest (const char *path="networks/models.json")
 Load the model manifest from a JSON file. More...
 
bool LoadModelManifest (nlohmann::json &models, const char *path="networks/models.json")
 Load the model manifest from a JSON file. More...
 

Detailed Description

Utilities for automatically downloading pre-trained models.

Macro Definition Documentation

◆ JSON_STR

#define JSON_STR (   x)    (x.is_string() ? x.get<std::string>() : "")

Helper macro for getting a string from a JSON element.

◆ JSON_STR_DEFAULT

#define JSON_STR_DEFAULT (   x,
  default 
)    (x.is_string() ? x.get<std::string>() : default)

Helper macro for getting a string from a JSON element.

Function Documentation

◆ DownloadModel() [1/2]

bool DownloadModel ( const char *  type,
const char *  name,
nlohmann::json &  model,
uint32_t  retries = 2 
)

Download a pre-trained model and return its JSON config.

◆ DownloadModel() [2/2]

bool DownloadModel ( nlohmann::json &  model,
uint32_t  retries = 2 
)

Download a pre-trained model given its JSON config.

◆ FindModel() [1/3]

bool FindModel ( const char *  type,
const char *  name 
)

Check if a model can be found in the model manifest.

◆ FindModel() [2/3]

bool FindModel ( const char *  type,
const char *  name,
nlohmann::json &  model 
)

Find a pre-trained model's config in the model manifest.

◆ FindModel() [3/3]

bool FindModel ( const char *  type,
const char *  name,
nlohmann::json &  models,
nlohmann::json &  model 
)

Find a pre-trained model's config in the model manifest.

◆ LoadModelManifest() [1/2]

bool LoadModelManifest ( const char *  path = "networks/models.json")

Load the model manifest from a JSON file.

◆ LoadModelManifest() [2/2]

bool LoadModelManifest ( nlohmann::json &  models,
const char *  path = "networks/models.json" 
)

Load the model manifest from a JSON file.