اپلیکیشن چند پلتفرمی دسته بندی تصاویر با فلاتر و Fastai — از صفر تا صد

1import aiohttp
2import asyncio
3import uvicorn
4from fastai import *
5from fastai.vision import *
6from io import BytesIO
7from starlette.applications import Starlette
8from starlette.middleware.cors import CORSMiddleware
9from starlette.responses import HTMLResponse, JSONResponse
10from starlette.staticfiles import StaticFiles
11
12
13export_file_url = 'DOWNLOAD_URL' #This will be the download url for your export.pkl file that you have already saved.
14export_file_name = 'export.pkl'
15
16classes = ['sweep', 'coverdrive', 'straightdrive', 'helicopter', 'scoop', 'pull'] #Replace these classes with the labels that your model identifies.
17path = Path(__file__).parent
18
19app = Starlette()
20app.add_middleware(CORSMiddleware, allow_origins=['*'], allow_headers=['X-Requested-With', 'Content-Type'])
21app.mount('/static', StaticFiles(directory='app/static'))
22
23
24async def download_file(url, dest):
25    if dest.exists(): return
26    async with aiohttp.ClientSession() as session:
27        async with session.get(url) as response:
28            data = await response.read()
29            with open(dest, 'wb') as f:
30                f.write(data)
31
32
33async def setup_learner():
34    await download_file(export_file_url, path / export_file_name)
35    try:
36        learn = load_learner(path, export_file_name)
37        return learn
38    except RuntimeError as e:
39        if len(e.args) > 0 and 'CPU-only machine' in e.args[0]:
40            print(e)
41            message = "\n\nThis model was trained with an old version of fastai and will not work in a CPU environment.\n\nPlease update the fastai library in your training environment and export your model again.\n\nSee instructions for 'Returning to work' at https://course.fast.ai."
42            raise RuntimeError(message)
43        else:
44            raise
45
46
47loop = asyncio.get_event_loop()
48tasks = [asyncio.ensure_future(setup_learner())]
49learn = loop.run_until_complete(asyncio.gather(*tasks))[0]
50loop.close()
51
52
53@app.route('/')
54async def homepage(request):
55    html_file = path / 'view' / 'index.html'
56    return HTMLResponse(html_file.open().read())
57
58
59@app.route('/analyze', methods=['POST'])
60async def analyze(request):
61    img_data = await request.form()
62    img_bytes = await (img_data['file'].read())
63    img = open_image(BytesIO(img_bytes))
64    prediction = learn.predict(img)[0]
65    return JSONResponse({'result': str(prediction)})
66
67
68
69if __name__ == '__main__':
70    if 'serve' in sys.argv:
71        uvicorn.run(app=app, host='0.0.0.0', port=5000, log_level="info")

1export_file_url = 'DOWNLOAD_URL'#Download_url for the .pkl file
2export_file_name = 'export.pkl'
3classes = ['sweep', 'coverdrive', 'straightdrive', 'helicopter', 'scoop', 'pull']

1app = Starlette()
2app.add_middleware(CORSMiddleware, allow_origins=['*'], allow_headers=['X-Requested-With', 'Content-Type'])
3app.mount('/static', StaticFiles(directory='app/static'))
4
5
6async def download_file(url, dest):
7    if dest.exists(): return
8    async with aiohttp.ClientSession() as session:
9        async with session.get(url) as response:
10            data = await response.read()
11            with open(dest, 'wb') as f:
12                f.write(data)
13
14
15async def setup_learner():
16    await download_file(export_file_url, path / export_file_name)
17    try:
18        learn = load_learner(path, export_file_name)
19        return learn
20    except RuntimeError as e:
21        if len(e.args) > 0 and 'CPU-only machine' in e.args[0]:
22            print(e)
23            message = "\n\nThis model was trained with an old version of fastai and will not work in a CPU environment.\n\nPlease update the fastai library in your training environment and export your model again.\n\nSee instructions for 'Returning to work' at https://course.fast.ai."
24            raise RuntimeError(message)
25        else:
26            raise

1loop = asyncio.get_event_loop()
2tasks = [asyncio.ensure_future(setup_learner())]
3learn = loop.run_until_complete(asyncio.gather(*tasks))[0]
4loop.close()
5
6
7@app.route('/')
8async def homepage(request):
9    html_file = path / 'view' / 'index.html'
10    return HTMLResponse(html_file.open().read())
11
12
13@app.route('/analyze', methods=['POST'])
14async def analyze(request):
15    img_data = await request.form()
16    img_bytes = await (img_data['file'].read())
17    img = open_image(BytesIO(img_bytes))
18    prediction = learn.predict(img)[0]
19    return JSONResponse({'result': str(prediction)})

1import 'dart:convert';
2import 'package:image_picker/image_picker.dart';
3import 'package:flutter/material.dart';
4import 'package:http/http.dart' as http;
5import 'dart:io';
6import 'package:path/path.dart';
7import 'package:async/async.dart';
8
9String txt = "";
10String txt1 = "Upload or take an image to figure out what type of cricket shot it is";
11void main() {
12  runApp(new MaterialApp(
13    debugShowCheckedModeBanner: false,
14    title: "Cricket Shot Classifier",
15    home: new MyApp(),
16  ));
17}
18
19class MyApp extends StatefulWidget {
20  @override
21  _MyAppState createState() => _MyAppState();
22}
23
24class _MyAppState extends State<MyApp> {
25  File img;
26
27  // The fuction which will upload the image as a file
28  void upload(File imageFile) async {
29    var stream =
30    new http.ByteStream(DelegatingStream.typed(imageFile.openRead()));
31    var length = await imageFile.length();
32
33    String base =
34        "https://types-of-cricket-shots.onrender.com";
35
36    var uri = Uri.parse(base + '/analyze');
37
38    var request = new http.MultipartRequest("POST", uri);
39    var multipartFile = new http.MultipartFile('file', stream, length,
40        filename: basename(imageFile.path));
41    //contentType: new MediaType('image', 'png'));
42
43    request.files.add(multipartFile);
44    var response = await request.send();
45    print(response.statusCode);
46    response.stream.transform(utf8.decoder).listen((value) {
47      print(value);
48      int l = value.length;
49      txt = value;
50
51      setState(() {});
52    });
53  }
54
55
56  void image_picker(int a) async {
57    txt1 = "";
58    setState(() {
59
60    });
61    debugPrint("Image Picker Activated");
62    if (a == 0){
63      img = await ImagePicker.pickImage(source: ImageSource.camera);
64    }
65    else{
66      img = await ImagePicker.pickImage(source: ImageSource.gallery);
67    }
68
69    txt = "Analysing...";
70    debugPrint(img.toString());
71    upload(img);
72    setState(() {});
73  }
74
75
76
77  @override
78  Widget build(BuildContext context) {
79    return Scaffold(
80      appBar: new AppBar(
81        centerTitle: true,
82        title: new Text("Cricket Shot Classifier"),
83      ),
84      body: new Container(
85        child: Center(
86          child: Column(
87            children: <Widget>[
88              img == null
89                  ? new Text(
90                txt1,
91                style: TextStyle(
92                  fontWeight: FontWeight.bold,
93                  fontSize: 32.0,
94                ),
95              )
96                  : new Image.file(img,
97                  height: MediaQuery.of(context).size.height * 0.6,
98                  width: MediaQuery.of(context).size.width * 0.8),
99              new Text(
100                txt,
101                style: TextStyle(
102                  fontWeight: FontWeight.bold,
103                  fontSize: 32.0,
104                ),
105              )
106            ],
107          ),
108        ),
109      ),
110    floatingActionButton: new Stack(
111      children: <Widget>[
112        Align(
113            alignment: Alignment(1.0, 1.0),
114            child: new FloatingActionButton(
115            onPressed: (){
116              image_picker(0);
117            },
118            child: new Icon(Icons.camera_alt),
119          )
120        ),
121        Align(
122          alignment: Alignment(1.0, 0.8),
123          child: new FloatingActionButton(
124              onPressed: (){
125                image_picker(1);
126              },
127              child: new Icon(Icons.file_upload)
128          )
129        ),
130      ],
131    ),
132    );
133  }
134}