object_detector 0.5.0

Object detection using ORT and the yoloe-26-seg model. This model can detect multiple objects per image, each having a tag, pixel-level mask, and a boundingbox. It's pretrained, it has a vocabulary of 4000+ objects.
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

use crate::ObjectDetectorError;
use crate::structs::{DetectorType, ModelScale};
use hf_hub::api::tokio::Api;

/// Details for fetching model files from `HuggingFace` Hub.
pub struct HfModel {
    /// Repository ID (e.g., "user/repo")
    pub id: String,
    /// Filename within the repository
    pub file: String,
}

impl HfModel {
    pub const DEFAULT_REPO_ID: &'static str = "RuteNL/yolo26-object-detection-ONNX";
    pub const DEFAULT_CLIP_REPO: &'static str = "RuteNL/MobileCLIP2-B-OpenCLIP-ONNX";

    #[must_use]
    pub fn get_model_file_path(
        detector_type: DetectorType,
        scale: ModelScale,
        include_mask: bool,
    ) -> String {
        let folder = match detector_type {
            #[cfg(feature = "promptable")]
            DetectorType::Promptable => "promptable",
            DetectorType::PromptFree => "prompt_free",
            #[allow(unreachable_patterns)]
            _ => "prompt_free",
        };
        let type_string = match detector_type {
            #[cfg(feature = "promptable")]
            DetectorType::Promptable => "promptable",
            DetectorType::PromptFree => "pf",
            #[allow(unreachable_patterns)]
            _ => "pf",
        };
        let scale_string = match scale {
            ModelScale::Nano => "n",
            ModelScale::Small => "s",
            ModelScale::Medium => "m",
            ModelScale::Large => "l",
            ModelScale::XLarge => "x",
        };
        let mask_string = if include_mask { "seg" } else { "det" };
        format!("{folder}/yoloe-26{scale_string}-{mask_string}-{type_string}.onnx")
    }

    #[must_use]
    pub fn default_prompt_free() -> Self {
        Self {
            id: Self::DEFAULT_REPO_ID.to_owned(),
            file: Self::get_model_file_path(DetectorType::PromptFree, ModelScale::Large, true),
        }
    }

    #[must_use]
    pub fn default_prompt_free_data() -> Self {
        Self {
            id: Self::DEFAULT_REPO_ID.to_owned(),
            file: format!(
                "{}.data",
                Self::get_model_file_path(DetectorType::PromptFree, ModelScale::Large, true)
            ),
        }
    }

    #[must_use]
    pub fn default_vocabulary() -> Self {
        Self {
            id: Self::DEFAULT_REPO_ID.to_owned(),
            file: "prompt_free/vocabulary_4585.json".to_owned(),
        }
    }

    #[cfg(feature = "promptable")]
    #[must_use]
    pub fn default_promptable() -> Self {
        Self {
            id: Self::DEFAULT_REPO_ID.to_owned(),
            file: Self::get_model_file_path(DetectorType::Promptable, ModelScale::Large, true),
        }
    }

    #[cfg(feature = "promptable")]
    #[must_use]
    pub fn default_promptable_data() -> Self {
        Self {
            id: Self::DEFAULT_REPO_ID.to_owned(),
            file: format!(
                "{}.data",
                Self::get_model_file_path(DetectorType::Promptable, ModelScale::Large, true)
            ),
        }
    }

    #[cfg(feature = "promptable")]
    #[must_use]
    pub fn default_clip_embedder() -> String {
        Self::DEFAULT_CLIP_REPO.to_owned()
    }
}

/// Downloads a file from `HuggingFace` Hub using the provided configuration.
pub async fn get_hf_model(model: HfModel) -> Result<std::path::PathBuf, ObjectDetectorError> {
    let api = Api::new()?;
    let repo = api.model(model.id);
    Ok(repo.get(&model.file).await?)
}