| name: latin_g2_jpqd |
| description: Latin script text recognition model (CRNN) optimized with JPQD quantization |
| framework: ONNX |
| task: text-recognition |
| domain: computer-vision |
| subdomain: optical-character-recognition |
|
|
| model_info: |
| architecture: CRNN (CNN + BiLSTM + CTC) |
| language: Latin script languages |
| supported_languages: |
| - English |
| - Spanish |
| - French |
| - German |
| - Italian |
| - Portuguese |
| - Dutch |
| - Polish |
| - Czech |
| - Romanian |
| - And other Latin-based languages |
| num_classes: 352 |
| original_source: EasyOCR |
| optimization: JPQD quantization |
| |
| specifications: |
| input_shape: [1, 1, 32, 100] |
| input_type: float32 |
| input_format: Grayscale |
| output_shape: [1, 25, 352] |
| output_type: float32 |
| batch_size: dynamic |
| sequence_length: 25 |
| |
| performance: |
| original_size_mb: 14.7 |
| optimized_size_mb: 8.5 |
| compression_ratio: 3.97 |
| inference_time_cpu_ms: ~12 |
| accuracy_retention: ">95%" |
| |
| deployment: |
| runtime: onnxruntime |
| hardware: CPU-optimized |
| precision: INT8 weights, FP32 activations |
| memory_usage_mb: ~15 |
| |
| usage: |
| preprocessing: |
| - Convert to grayscale |
| - Resize to 32x100 (height x width) |
| - Normalize to [0,1] |
| - Add batch and channel dimensions |
| postprocessing: |
| - Apply CTC decoding |
| - Convert indices to characters |
| - Remove blank tokens and duplicates |
|
|
| supported_characters: |
| basic_latin: "a-z, A-Z, 0-9" |
| latin_extended: "À-ÿ (Latin-1 Supplement)" |
| punctuation: "Standard punctuation marks" |
| symbols: "Common symbols and currency" |
| diacritics: "Accented characters for European languages" |
| |
| character_coverage: |
| - "Basic Latin (U+0020-U+007F)" |
| - "Latin-1 Supplement (U+0080-U+00FF)" |
| - "Latin Extended-A (U+0100-U+017F)" |
| - "Latin Extended-B (U+0180-U+024F)" |
| - "Combining Diacritical Marks (U+0300-U+036F)" |
| |
| training_data: |
| type: Multilingual synthetic and real text images |
| languages: Multiple Latin script languages |
| domains: Documents, natural scenes, printed text, handwriting |
| |
| use_cases: |
| - Multilingual document processing |
| - European language OCR |
| - International text recognition |
| - Multilingual forms processing |
|
|
| license: apache-2.0 |
| tags: |
| - text-recognition |
| - latin |
| - multilingual |
| - crnn |
| - lstm |
| - ocr |
| - onnx |
| - quantized |
| - jpqd |
| - european-languages |
