oar_ocr_core/domain/tasks/
table_classification.rs

1//! Concrete task implementations for table classification.
2//!
3//! This module provides the table classification task that classifies table images
4//! as either "wired_table" (tables with borders) or "wireless_table" (tables without borders).
5
6use super::document_orientation::Classification;
7use super::validation::ensure_non_empty_images;
8use crate::ConfigValidator;
9use crate::core::OCRError;
10use crate::core::traits::TaskDefinition;
11use crate::core::traits::task::{ImageTaskInput, Task, TaskType};
12use crate::utils::ScoreValidator;
13use serde::{Deserialize, Serialize};
14
15/// Configuration for table classification task.
16#[derive(Debug, Clone, Serialize, Deserialize, ConfigValidator)]
17pub struct TableClassificationConfig {
18    /// Score threshold for classification (default: 0.5)
19    #[validate(range(min = 0.0, max = 1.0))]
20    pub score_threshold: f32,
21    /// Number of top predictions to return (default: 2)
22    #[validate(min = 1)]
23    pub topk: usize,
24}
25
26impl Default for TableClassificationConfig {
27    fn default() -> Self {
28        Self {
29            score_threshold: 0.5,
30            topk: 2,
31        }
32    }
33}
34
35/// Output from table classification task.
36#[derive(Debug, Clone)]
37pub struct TableClassificationOutput {
38    /// Classification results per image
39    pub classifications: Vec<Vec<Classification>>,
40}
41
42impl TableClassificationOutput {
43    /// Creates an empty table classification output.
44    pub fn empty() -> Self {
45        Self {
46            classifications: Vec::new(),
47        }
48    }
49
50    /// Creates a table classification output with the given capacity.
51    pub fn with_capacity(capacity: usize) -> Self {
52        Self {
53            classifications: Vec::with_capacity(capacity),
54        }
55    }
56}
57
58impl TaskDefinition for TableClassificationOutput {
59    const TASK_NAME: &'static str = "table_classification";
60    const TASK_DOC: &'static str =
61        "Table classification - classifying table images as wired or wireless";
62
63    fn empty() -> Self {
64        TableClassificationOutput::empty()
65    }
66}
67
68/// Table classification task implementation.
69#[derive(Debug, Default)]
70pub struct TableClassificationTask {
71    _config: TableClassificationConfig,
72}
73
74impl TableClassificationTask {
75    /// Creates a new table classification task.
76    pub fn new(config: TableClassificationConfig) -> Self {
77        Self { _config: config }
78    }
79}
80
81impl Task for TableClassificationTask {
82    type Config = TableClassificationConfig;
83    type Input = ImageTaskInput;
84    type Output = TableClassificationOutput;
85
86    fn task_type(&self) -> TaskType {
87        TaskType::TableClassification
88    }
89
90    fn validate_input(&self, input: &Self::Input) -> Result<(), OCRError> {
91        ensure_non_empty_images(&input.images, "No images provided for table classification")?;
92
93        Ok(())
94    }
95
96    fn validate_output(&self, output: &Self::Output) -> Result<(), OCRError> {
97        let validator = ScoreValidator::new_unit_range("score");
98
99        for (idx, classifications) in output.classifications.iter().enumerate() {
100            for classification in classifications.iter() {
101                // Validate class IDs (should be 0-1 for 2 table types)
102                if classification.class_id > 1 {
103                    return Err(OCRError::InvalidInput {
104                        message: format!(
105                            "Image {}: invalid class_id {}. Expected 0-1 (wired_table, wireless_table)",
106                            idx, classification.class_id
107                        ),
108                    });
109                }
110            }
111
112            // Validate score ranges
113            let scores: Vec<f32> = classifications.iter().map(|c| c.score).collect();
114            validator.validate_scores_with(&scores, |class_idx| {
115                format!("Image {}, classification {}", idx, class_idx)
116            })?;
117        }
118
119        Ok(())
120    }
121
122    fn empty_output(&self) -> Self::Output {
123        TableClassificationOutput::empty()
124    }
125}
126
127#[cfg(test)]
128mod tests {
129    use super::*;
130    use image::RgbImage;
131
132    #[test]
133    fn test_table_classification_task_creation() {
134        let task = TableClassificationTask::default();
135        assert_eq!(task.task_type(), TaskType::TableClassification);
136    }
137
138    #[test]
139    fn test_input_validation() {
140        let task = TableClassificationTask::default();
141
142        // Empty images should fail
143        let empty_input = ImageTaskInput::new(vec![]);
144        assert!(task.validate_input(&empty_input).is_err());
145
146        // Valid images should pass
147        let valid_input = ImageTaskInput::new(vec![RgbImage::new(100, 100)]);
148        assert!(task.validate_input(&valid_input).is_ok());
149    }
150
151    #[test]
152    fn test_output_validation() {
153        let task = TableClassificationTask::default();
154
155        // Valid output should pass
156        let classification1 = Classification::new(0, "wired_table".to_string(), 0.85);
157        let classification2 = Classification::new(1, "wireless_table".to_string(), 0.15);
158        let output = TableClassificationOutput {
159            classifications: vec![vec![classification1, classification2]],
160        };
161        assert!(task.validate_output(&output).is_ok());
162
163        // Invalid class ID should fail (should be 0-1)
164        let bad_classification = Classification::new(2, "invalid".to_string(), 0.95);
165        let bad_output = TableClassificationOutput {
166            classifications: vec![vec![bad_classification]],
167        };
168        assert!(task.validate_output(&bad_output).is_err());
169
170        // Invalid score should fail
171        let bad_score_classification = Classification::new(0, "wired_table".to_string(), 1.5);
172        let bad_score_output = TableClassificationOutput {
173            classifications: vec![vec![bad_score_classification]],
174        };
175        assert!(task.validate_output(&bad_score_output).is_err());
176    }
177}
oar_ocr_core/domain/tasks/table_classification.rs

oar_ocr_core/domain/tasks/
table_classification.rs
