File size: 4,933 Bytes
7279c87
 
 
 
 
 
 
263280b
7279c87
 
 
 
 
263280b
 
7279c87
467ec0d
7279c87
 
 
263280b
 
 
 
467ec0d
263280b
 
 
 
 
 
 
7279c87
467ec0d
 
 
 
 
 
 
 
 
 
 
 
7279c87
 
 
 
 
263280b
 
 
 
467ec0d
7279c87
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
263280b
7279c87
 
 
263280b
7279c87
 
 
 
 
 
 
 
 
 
263280b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import os
import time
import tempfile
from pathlib import Path

from fastapi import FastAPI, File, UploadFile
from fastapi.responses import HTMLResponse, JSONResponse
from fastapi.staticfiles import StaticFiles

from inference import load_model, predict_video
from video_preprocess import PreprocessConfig

APP_DIR = Path(__file__).resolve().parent
UI_DIR = APP_DIR / "ui"
DIST_DIR = UI_DIR / "dist"

app = FastAPI(title="LumiSign Sign Inference")

MODEL = None
LABEL_MAP = None
DEFAULT_LOCAL_CHECKPOINT = APP_DIR / "transformer_large.pth"
DEFAULT_MODEL_CHECKPOINT = (
    str(DEFAULT_LOCAL_CHECKPOINT) if DEFAULT_LOCAL_CHECKPOINT.is_file() else None
)
MODEL_DATASET = os.getenv("MODEL_DATASET", "isl_split_dataset")
MODEL_TYPE = os.getenv("MODEL_TYPE", "transformer")
MODEL_TRANSFORMER_SIZE = os.getenv(
    "MODEL_TRANSFORMER_SIZE",
    "large" if DEFAULT_MODEL_CHECKPOINT else "small",
)
MODEL_CHECKPOINT = os.getenv("MODEL_CHECKPOINT", DEFAULT_MODEL_CHECKPOINT)
MODEL_MAX_FRAME_LEN = int(os.getenv("MODEL_MAX_FRAME_LEN", "169"))

DEFAULT_LABEL_MAP_CANDIDATES = [
    APP_DIR / "label_maps" / f"label_map_{MODEL_DATASET}.json",
    APP_DIR / "label_maps" / "label_map_isl_split_dataset.json",
    APP_DIR / "label_maps" / "label_map_islsplit.json",
]
MODEL_LABEL_MAP_PATH = os.getenv("MODEL_LABEL_MAP_PATH")
if not MODEL_LABEL_MAP_PATH:
    for candidate in DEFAULT_LABEL_MAP_CANDIDATES:
        if candidate.is_file():
            MODEL_LABEL_MAP_PATH = str(candidate)
            break


@app.on_event("startup")
def _load():
    global MODEL, LABEL_MAP
    MODEL, LABEL_MAP = load_model(
        dataset=MODEL_DATASET,
        model_type=MODEL_TYPE,
        transformer_size=MODEL_TRANSFORMER_SIZE,
        checkpoint_path=MODEL_CHECKPOINT,
        label_map_path=MODEL_LABEL_MAP_PATH,
    )


@app.post("/predict")
async def predict(file: UploadFile = File(...)):
    if MODEL is None or LABEL_MAP is None:
        return JSONResponse({"error": "Model not loaded"}, status_code=500)

    suffix = Path(file.filename).suffix if file.filename else ".mp4"
    with tempfile.NamedTemporaryFile(delete=False, suffix=suffix) as tmp:
        contents = await file.read()
        tmp.write(contents)
        tmp_path = tmp.name

    preprocess_config = PreprocessConfig(
        apply_darken=False,
        apply_brighten=True,
        darken_min=0.3,
        darken_max=0.8,
        brighten_method="clahe",
        brighten_gamma_min=1.2,
        brighten_gamma_max=1.8,
    )

    start = time.time()
    try:
        result = predict_video(
            tmp_path,
            dataset=MODEL_DATASET,
            model=MODEL,
            label_map=LABEL_MAP,
            preprocess_config=preprocess_config,
            max_frame_len=MODEL_MAX_FRAME_LEN,
        )
    finally:
        try:
            os.remove(tmp_path)
        except OSError:
            pass

    elapsed_ms = int((time.time() - start) * 1000)
    result["elapsed_ms"] = elapsed_ms
    return JSONResponse(result)


def _ui_built() -> bool:
    index_path = DIST_DIR / "index.html"
    if not index_path.is_file():
        return False

    src_dir = UI_DIR / "src"
    if not src_dir.is_dir():
        return True

    latest_src_mtime = 0.0
    for root, _, files in os.walk(src_dir):
        for name in files:
            try:
                latest_src_mtime = max(
                    latest_src_mtime,
                    (Path(root) / name).stat().st_mtime,
                )
            except OSError:
                continue

    try:
        return index_path.stat().st_mtime >= latest_src_mtime
    except OSError:
        return True


def _configure_ui(app: FastAPI) -> None:
    if _ui_built():
        app.mount("/", StaticFiles(directory=DIST_DIR, html=True), name="ui")
        return

    @app.get("/")
    def index():
        ui_path = str(UI_DIR)
        html = f"""
        <!doctype html>
        <html lang="en">
          <head>
            <meta charset="utf-8" />
            <meta name="viewport" content="width=device-width, initial-scale=1" />
            <title>UI Not Built</title>
            <style>
              body {{ font-family: Arial, sans-serif; padding: 32px; background: #f8fafc; color: #0f172a; }}
              pre {{ background: #0f172a; color: #e2e8f0; padding: 12px; border-radius: 8px; }}
              code {{ font-family: monospace; }}
            </style>
          </head>
          <body>
            <h1>UI not built yet</h1>
            <p>To run the React UI in development:</p>
            <pre>cd {ui_path}
npm install
npm run dev</pre>
            <p>Then open <code>http://localhost:5173</code> (it proxies <code>/predict</code>).</p>
            <p>To serve the built UI from FastAPI:</p>
            <pre>cd {ui_path}
npm run build</pre>
            <p>Restart the backend after building.</p>
          </body>
        </html>
        """
        return HTMLResponse(html)


_configure_ui(app)