deepinsight/insightface · error · ValueError

Model ' ' not found. Available models

Error message

Model '{name}' not found. Available models: {list(self._MODEL_LIST.keys())}

What it means

In the legacy (non-ModelScope) download path, get_model looks up name in the static _MODEL_LIST registry; an unknown key raises ValueError listing all valid model names. This is pure input validation before any file or network access happens.

Solutions

1. Use one of the names printed in the error message itself (list(self._MODEL_LIST.keys()))
2. Print the registry to discover valid names: inspect resource._MODEL_LIST at runtime
3. If you need a newer/renamed model, upgrade the inspireface package

Example fix

# before
path = resource.get_model('Pikachu')
# after
valid = list(resource._MODEL_LIST)
print(valid)  # see exact keys
path = resource.get_model(valid[0])

valid = list(resource._MODEL_LIST.keys())
assert name in valid, f'{name!r} not in {valid}'

def is_known_model(resource, name: str) -> bool:
    return name in resource._MODEL_LIST

try:
    path = resource.get_model(name)
except ValueError:
    name = sorted(resource._MODEL_LIST)[0]  # pick a valid model
    path = resource.get_model(name)

When it happens

Trigger: Calling resource.get_model(name) in OSS mode with a string that isn't a key of _MODEL_LIST (typo, wrong casing, or a model removed/renamed in this version).

Common situations: Upgrading inspireface and re-running old scripts whose model names changed; guessing names like 'Pikachu' vs 'pikachu'; copy-pasting names from outdated docs.

Related errors

• exec_param must be SessionCustomParameter or int
• HERR_INVALID_FACE_FEATURE
• HERR_INVALID_PARAM
• Image must be numpy.ndarray or ImageStream
• Failed to download model

AI-assisted analysis of deepinsight/insightface@7fadd420c2 (2026-08-28). Data as JSON: /api/errors/04eb7b5fbaad6290. Report an issue: GitHub.

Appendix: source

Thrown at cpp-package/inspireface/python/inspireface/modules/utils/resource.py:163

        Args:
            name: Model name
            re_download: Force re-download if True
            ignore_verification: Skip model hash verification if True
            
        Returns:
            str: Full path to model file
        """
        # Check global OSS setting first, then instance setting
        use_oss = USE_OSS_DOWNLOAD
        use_modelscope_actual = self.use_modelscope and not use_oss
        
        # Use ModelScope download if enabled and OSS is not forced
        if use_modelscope_actual:
            return self._download_from_modelscope_with_cache(name, re_download)
            
        # Original download logic for backwards compatibility
        if name not in self._MODEL_LIST:
            raise ValueError(f"Model '{name}' not found. Available models: {list(self._MODEL_LIST.keys())}")

        model_info = self._MODEL_LIST[name]
        model_file = self.models_dir / model_info["filename"]
        downloading_flag = model_file.with_suffix('.downloading')
        
        # Check if model exists and is complete
        if model_file.exists() and not downloading_flag.exists() and not re_download:
            if ignore_verification:
                print(f"Warning: Model verification skipped for '{name}' as requested.")
                return str(model_file)
                
            current_hash = get_file_hash_sha256(model_file)
            if current_hash == model_info["md5"]:
                return str(model_file)
            else:
                print(f"Model file hash mismatch for '{name}'. Re-downloading...")

        # Start download

View on GitHub (pinned to 7fadd420c2)