yscv 0.1.8

Rust computer vision and deep learning framework
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200

#![doc = include_str!("../README.md")]

pub use yscv_autograd as autograd;
pub use yscv_cli as cli;
pub use yscv_detect as detect;
pub use yscv_eval as eval;
pub use yscv_imgproc as imgproc;
pub use yscv_kernels as kernels;
pub use yscv_model as model;
pub use yscv_onnx as onnx;
pub use yscv_optim as optim;
pub use yscv_recognize as recognize;
pub use yscv_tensor as tensor;
pub use yscv_track as track;
pub use yscv_video as video;

/// Commonly used types from across the yscv ecosystem.
pub mod prelude {
    // tensor
    pub use yscv_tensor::{DType, Device, Tensor, TensorError};

    // autograd
    pub use yscv_autograd::{Graph, NodeId};

    // model
    pub use yscv_model::{
        Compose, DataLoader, DataLoaderConfig, InferencePipeline, ModelLayer, SequentialModel,
        Trainer, TrainerConfig, Transform,
    };

    // losses
    pub use yscv_model::{
        bce_loss, cross_entropy_loss, dice_loss, focal_loss, mse_loss, smooth_l1_loss,
    };

    // callbacks
    pub use yscv_model::{BestModelCheckpoint, EarlyStopping};

    // transforms
    pub use yscv_model::{Normalize, PermuteDims, ScaleValues};

    // config types
    pub use yscv_model::{LossKind, OptimizerKind, SupervisedDataset};

    // optim
    pub use yscv_optim::{Adam, LearningRate, Sgd};

    // imgproc — module re-export plus essential functions and types
    pub use crate::imgproc;
    pub use yscv_imgproc::{
        ImageF32, ImageU8, ImgProcError, imread, imread_gray, imwrite, rgb_to_grayscale,
    };

    // onnx — model loading, inference, and error type
    pub use yscv_onnx::{
        OnnxError, OnnxModel, load_onnx_model, load_onnx_model_from_file, run_onnx_model,
    };

    // detect
    pub use yscv_detect::BoundingBox;

    // eval metrics
    pub use yscv_eval::{accuracy, confusion_matrix, f1_score};
}

#[cfg(test)]
mod tests {
    use super::prelude::*;

    #[test]
    fn smoke_create_tensor() {
        let t = Tensor::from_vec(vec![2, 3], vec![1.0, 2.0, 3.0, 4.0, 5.0, 6.0]).unwrap();
        assert_eq!(t.shape(), &[2, 3]);
    }

    #[test]
    fn smoke_from_slice() {
        let t = Tensor::from_slice(&[1.0, 2.0, 3.0]);
        assert_eq!(t.shape(), &[3]);
        assert_eq!(t.data(), &[1.0, 2.0, 3.0]);
    }

    #[test]
    fn smoke_full() {
        let t = Tensor::full(vec![2, 3], 7.0).unwrap();
        assert_eq!(t.shape(), &[2, 3]);
        assert!(t.data().iter().all(|&v| v == 7.0));
    }

    #[test]
    fn smoke_display() {
        let t = Tensor::from_slice(&[1.0, 2.0, 3.0]);
        let s = format!("{t}");
        assert!(s.contains("Tensor("));
        assert!(s.contains("[3]"));
    }

    #[test]
    fn smoke_operator_add() {
        let a = Tensor::from_slice(&[1.0, 2.0]);
        let b = Tensor::from_slice(&[3.0, 4.0]);
        let c = &a + &b;
        assert_eq!(c.data(), &[4.0, 6.0]);
    }

    #[test]
    fn smoke_operator_sub() {
        let a = Tensor::from_slice(&[5.0, 3.0]);
        let b = Tensor::from_slice(&[1.0, 1.0]);
        let c = &a - &b;
        assert_eq!(c.data(), &[4.0, 2.0]);
    }

    #[test]
    fn smoke_operator_mul_scalar() {
        let a = Tensor::from_slice(&[2.0, 3.0]);
        let b = &a * 10.0;
        assert_eq!(b.data(), &[20.0, 30.0]);
    }

    #[test]
    fn smoke_operator_neg() {
        let a = Tensor::from_slice(&[1.0, -2.0]);
        let b = -&a;
        assert_eq!(b.data(), &[-1.0, 2.0]);
    }

    #[test]
    fn smoke_prelude_types_accessible() {
        // Verify new prelude types compile.
        let _: Option<DType> = None;
        let _: Option<Device> = None;
        let _: Option<ImgProcError> = None;
        let _: Option<OnnxError> = None;
        let _: Option<OnnxModel> = None;
    }

    #[test]
    fn smoke_build_sequential_model() {
        let graph = Graph::new();
        let model = SequentialModel::new(&graph);
        assert_eq!(model.layers().len(), 0);
    }

    #[test]
    fn smoke_inference_pipeline() {
        let graph = Graph::new();
        let model = SequentialModel::new(&graph);
        let pipeline = InferencePipeline::new(model);
        let input = Tensor::from_vec(vec![1, 4], vec![1.0, 2.0, 3.0, 4.0]).unwrap();
        // Empty model returns input unchanged.
        let output = pipeline.run(&input).unwrap();
        assert_eq!(output.shape(), &[1, 4]);
    }

    #[test]
    fn smoke_prelude_new_imports() {
        // Verify loss functions are callable via the prelude.
        let mut graph = Graph::new();
        let pred_t = Tensor::from_vec(vec![3], vec![0.9, 0.1, 0.0]).unwrap();
        let targ_t = Tensor::from_vec(vec![3], vec![1.0, 0.0, 0.0]).unwrap();
        let pred_node = graph.constant(pred_t);
        let targ_node = graph.constant(targ_t);
        let loss_node = mse_loss(&mut graph, pred_node, targ_node).unwrap();
        let loss_val = graph.value(loss_node).unwrap().data()[0];
        assert!(loss_val >= 0.0);

        // Verify eval metrics are callable via the prelude.
        let preds: Vec<usize> = vec![0, 1, 2, 0];
        let labels: Vec<usize> = vec![0, 1, 2, 1];
        let acc = accuracy(&preds, &labels).unwrap();
        assert!((acc - 0.75).abs() < 1e-5);

        // Verify transform types are accessible.
        let _scale = ScaleValues::new(1.0 / 255.0);
        let _norm = Normalize::new(vec![0.5], vec![0.5]);
        let _perm = PermuteDims::new(vec![2, 0, 1]);

        // Verify config types are accessible (just reference the types).
        let _: Option<OptimizerKind> = None;
        let _: Option<LossKind> = None;
    }

    #[test]
    fn smoke_compose_transform() {
        use yscv_model::{Normalize, ScaleValues};

        let compose = Compose::new()
            .add(ScaleValues::new(1.0 / 255.0))
            .add(Normalize::new(vec![0.5], vec![0.5]));
        let input = Tensor::from_vec(vec![2], vec![255.0, 0.0]).unwrap();
        let output = compose.apply(&input).unwrap();
        assert_eq!(output.shape(), &[2]);
        // 255 * (1/255) = 1.0, then (1.0 - 0.5) / 0.5 = 1.0
        let data = output.data();
        assert!((data[0] - 1.0).abs() < 1e-5);
        // 0 * (1/255) = 0.0, then (0.0 - 0.5) / 0.5 = -1.0
        assert!((data[1] - (-1.0)).abs() < 1e-5);
    }
}