sync

.gitattributes

@@ -1,35 +1,34 @@
 *.7z filter=lfs diff=lfs merge=lfs -text
 *.arrow filter=lfs diff=lfs merge=lfs -text
 *.bin filter=lfs diff=lfs merge=lfs -text
+*.bin.* filter=lfs diff=lfs merge=lfs -text
 *.bz2 filter=lfs diff=lfs merge=lfs -text
-*.ckpt filter=lfs diff=lfs merge=lfs -text
 *.ftz filter=lfs diff=lfs merge=lfs -text
 *.gz filter=lfs diff=lfs merge=lfs -text
 *.h5 filter=lfs diff=lfs merge=lfs -text
 *.joblib filter=lfs diff=lfs merge=lfs -text
 *.lfs.* filter=lfs diff=lfs merge=lfs -text
-*.mlmodel filter=lfs diff=lfs merge=lfs -text
 *.model filter=lfs diff=lfs merge=lfs -text
 *.msgpack filter=lfs diff=lfs merge=lfs -text
-*.npy filter=lfs diff=lfs merge=lfs -text
-*.npz filter=lfs diff=lfs merge=lfs -text
 *.onnx filter=lfs diff=lfs merge=lfs -text
 *.ot filter=lfs diff=lfs merge=lfs -text
 *.parquet filter=lfs diff=lfs merge=lfs -text
 *.pb filter=lfs diff=lfs merge=lfs -text
-*.pickle filter=lfs diff=lfs merge=lfs -text
-*.pkl filter=lfs diff=lfs merge=lfs -text
 *.pt filter=lfs diff=lfs merge=lfs -text
 *.pth filter=lfs diff=lfs merge=lfs -text
 *.rar filter=lfs diff=lfs merge=lfs -text
-*.safetensors filter=lfs diff=lfs merge=lfs -text
 saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.tar.* filter=lfs diff=lfs merge=lfs -text
-*.tar filter=lfs diff=lfs merge=lfs -text
 *.tflite filter=lfs diff=lfs merge=lfs -text
 *.tgz filter=lfs diff=lfs merge=lfs -text
-*.wasm filter=lfs diff=lfs merge=lfs -text
 *.xz filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
-*.zst filter=lfs diff=lfs merge=lfs -text
-*tfevents* filter=lfs diff=lfs merge=lfs -text
+*.zstandard filter=lfs diff=lfs merge=lfs -text
+*.tfevents* filter=lfs diff=lfs merge=lfs -text
+*.db* filter=lfs diff=lfs merge=lfs -text
+*.ark* filter=lfs diff=lfs merge=lfs -text
+**/*ckpt*data* filter=lfs diff=lfs merge=lfs -text
+**/*ckpt*.meta filter=lfs diff=lfs merge=lfs -text
+**/*ckpt*.index filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1,4 @@
+__pycache__/*
+*.pth
+flagged/*
+rename.sh

README.md

@@ -1,13 +1,13 @@
 ---
 title: SVHN Recognition
-emoji: ⚡
+emoji: 🚪
 colorFrom: yellow
 colorTo: purple
 sdk: gradio
-sdk_version: 4.12.0
+sdk_version: 4.36.0
 app_file: app.py
 pinned: false
 license: mit
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+The Doorplate Recognition model is implemented using a deep convolutional neural network in PyTorch, with the objective of discerning multi-digit doorplate numbers from street view images. Utilizing the SVHN dataset extracted from Google Street View house numbers, the model is trained to identify sets of Arabic digits (0-9) within each image. The PyTorch implementation exhibits a commendable level of accuracy, achieving a tested precision of up to 89%. When users upload images containing doorplate numbers and submit them, the system yields precise recognition results for the digits present in the doorplate. This implementation provides a robust and user-friendly solution for doorplate number identification, demonstrating practical applications in the realm of image-based digit recognition.

app.py

@@ -0,0 +1,101 @@
+import os
+import torch
+import random
+import warnings
+import gradio as gr
+from PIL import Image
+from model import Model
+from torchvision import transforms
+from modelscope import snapshot_download
+
+
+MODEL_DIR = snapshot_download("MuGeminorum/svhn", cache_dir="./__pycache__")
+
+
+def infer(input_img: str, checkpoint_file: str):
+    try:
+        model = Model()
+        model.restore(f"{MODEL_DIR}/{checkpoint_file}")
+        outstr = ""
+        with torch.no_grad():
+            transform = transforms.Compose(
+                [
+                    transforms.Resize([64, 64]),
+                    transforms.CenterCrop([54, 54]),
+                    transforms.ToTensor(),
+                    transforms.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5]),
+                ]
+            )
+            image = Image.open(input_img)
+            image = image.convert("RGB")
+            image = transform(image)
+            images = image.unsqueeze(dim=0)
+            (
+                length_logits,
+                digit1_logits,
+                digit2_logits,
+                digit3_logits,
+                digit4_logits,
+                digit5_logits,
+            ) = model.eval()(images)
+            length_prediction = length_logits.max(1)[1]
+            digit1_prediction = digit1_logits.max(1)[1]
+            digit2_prediction = digit2_logits.max(1)[1]
+            digit3_prediction = digit3_logits.max(1)[1]
+            digit4_prediction = digit4_logits.max(1)[1]
+            digit5_prediction = digit5_logits.max(1)[1]
+            output = [
+                digit1_prediction.item(),
+                digit2_prediction.item(),
+                digit3_prediction.item(),
+                digit4_prediction.item(),
+                digit5_prediction.item(),
+            ]
+
+            for i in range(length_prediction.item()):
+                outstr += str(output[i])
+
+        return outstr
+
+    except Exception as e:
+        return f"{e}"
+
+
+def get_files(dir_path=MODEL_DIR, ext=".pth"):
+    files_and_folders = os.listdir(dir_path)
+    outputs = []
+    for file in files_and_folders:
+        if file.endswith(ext):
+            outputs.append(file)
+
+    return outputs
+
+
+if __name__ == "__main__":
+    warnings.filterwarnings("ignore")
+    models = get_files()
+    images = get_files(f"{MODEL_DIR}/examples", ".png")
+    samples = []
+    for img in images:
+        samples.append(
+            [
+                f"{MODEL_DIR}/examples/{img}",
+                models[random.randint(0, len(models) - 1)],
+            ]
+        )
+
+    gr.Interface(
+        fn=infer,
+        inputs=[
+            gr.Image(label="上传图片 Upload an image", type="filepath"),
+            gr.Dropdown(
+                label="选择权重 Select a model",
+                choices=models,
+                value=models[0],
+            ),
+        ],
+        outputs=gr.Textbox(label="识别结果 Recognition result", show_copy_button=True),
+        examples=samples,
+        allow_flagging="never",
+        cache_examples=False,
+    ).launch()

model.py

@@ -0,0 +1,158 @@
+import os
+import glob
+import torch
+import torch.jit
+import torch.nn as nn
+
+
+class Model(torch.jit.ScriptModule):
+    CHECKPOINT_FILENAME_PATTERN = "model-{}.pth"
+
+    __constants__ = [
+        "_hidden1",
+        "_hidden2",
+        "_hidden3",
+        "_hidden4",
+        "_hidden5",
+        "_hidden6",
+        "_hidden7",
+        "_hidden8",
+        "_hidden9",
+        "_hidden10",
+        "_features",
+        "_classifier",
+        "_digit_length",
+        "_digit1",
+        "_digit2",
+        "_digit3",
+        "_digit4",
+        "_digit5",
+    ]
+
+    def __init__(self):
+        super(Model, self).__init__()
+
+        self._hidden1 = nn.Sequential(
+            nn.Conv2d(in_channels=3, out_channels=48, kernel_size=5, padding=2),
+            nn.BatchNorm2d(num_features=48),
+            nn.ReLU(),
+            nn.MaxPool2d(kernel_size=2, stride=2, padding=1),
+            nn.Dropout(0.2),
+        )
+        self._hidden2 = nn.Sequential(
+            nn.Conv2d(in_channels=48, out_channels=64, kernel_size=5, padding=2),
+            nn.BatchNorm2d(num_features=64),
+            nn.ReLU(),
+            nn.MaxPool2d(kernel_size=2, stride=1, padding=1),
+            nn.Dropout(0.2),
+        )
+        self._hidden3 = nn.Sequential(
+            nn.Conv2d(in_channels=64, out_channels=128, kernel_size=5, padding=2),
+            nn.BatchNorm2d(num_features=128),
+            nn.ReLU(),
+            nn.MaxPool2d(kernel_size=2, stride=2, padding=1),
+            nn.Dropout(0.2),
+        )
+        self._hidden4 = nn.Sequential(
+            nn.Conv2d(in_channels=128, out_channels=160, kernel_size=5, padding=2),
+            nn.BatchNorm2d(num_features=160),
+            nn.ReLU(),
+            nn.MaxPool2d(kernel_size=2, stride=1, padding=1),
+            nn.Dropout(0.2),
+        )
+        self._hidden5 = nn.Sequential(
+            nn.Conv2d(in_channels=160, out_channels=192, kernel_size=5, padding=2),
+            nn.BatchNorm2d(num_features=192),
+            nn.ReLU(),
+            nn.MaxPool2d(kernel_size=2, stride=2, padding=1),
+            nn.Dropout(0.2),
+        )
+        self._hidden6 = nn.Sequential(
+            nn.Conv2d(in_channels=192, out_channels=192, kernel_size=5, padding=2),
+            nn.BatchNorm2d(num_features=192),
+            nn.ReLU(),
+            nn.MaxPool2d(kernel_size=2, stride=1, padding=1),
+            nn.Dropout(0.2),
+        )
+        self._hidden7 = nn.Sequential(
+            nn.Conv2d(in_channels=192, out_channels=192, kernel_size=5, padding=2),
+            nn.BatchNorm2d(num_features=192),
+            nn.ReLU(),
+            nn.MaxPool2d(kernel_size=2, stride=2, padding=1),
+            nn.Dropout(0.2),
+        )
+        self._hidden8 = nn.Sequential(
+            nn.Conv2d(in_channels=192, out_channels=192, kernel_size=5, padding=2),
+            nn.BatchNorm2d(num_features=192),
+            nn.ReLU(),
+            nn.MaxPool2d(kernel_size=2, stride=1, padding=1),
+            nn.Dropout(0.2),
+        )
+        self._hidden9 = nn.Sequential(nn.Linear(192 * 7 * 7, 3072), nn.ReLU())
+        self._hidden10 = nn.Sequential(nn.Linear(3072, 3072), nn.ReLU())
+
+        self._digit_length = nn.Sequential(nn.Linear(3072, 7))
+        self._digit1 = nn.Sequential(nn.Linear(3072, 11))
+        self._digit2 = nn.Sequential(nn.Linear(3072, 11))
+        self._digit3 = nn.Sequential(nn.Linear(3072, 11))
+        self._digit4 = nn.Sequential(nn.Linear(3072, 11))
+        self._digit5 = nn.Sequential(nn.Linear(3072, 11))
+
+    @torch.jit.script_method
+    def forward(self, x):
+        x = self._hidden1(x)
+        x = self._hidden2(x)
+        x = self._hidden3(x)
+        x = self._hidden4(x)
+        x = self._hidden5(x)
+        x = self._hidden6(x)
+        x = self._hidden7(x)
+        x = self._hidden8(x)
+        x = x.view(x.size(0), 192 * 7 * 7)
+        x = self._hidden9(x)
+        x = self._hidden10(x)
+
+        length_logits = self._digit_length(x)
+        digit1_logits = self._digit1(x)
+        digit2_logits = self._digit2(x)
+        digit3_logits = self._digit3(x)
+        digit4_logits = self._digit4(x)
+        digit5_logits = self._digit5(x)
+
+        return (
+            length_logits,
+            digit1_logits,
+            digit2_logits,
+            digit3_logits,
+            digit4_logits,
+            digit5_logits,
+        )
+
+    def store(self, path_to_dir, step, maximum=5):
+        path_to_models = glob.glob(
+            os.path.join(path_to_dir, Model.CHECKPOINT_FILENAME_PATTERN.format("*"))
+        )
+        if len(path_to_models) == maximum:
+            min_step = min(
+                [
+                    int(path_to_model.split("\\")[-1][6:-4])
+                    for path_to_model in path_to_models
+                ]
+            )
+            path_to_min_step_model = os.path.join(
+                path_to_dir, Model.CHECKPOINT_FILENAME_PATTERN.format(min_step)
+            )
+            os.remove(path_to_min_step_model)
+
+        path_to_checkpoint_file = os.path.join(
+            path_to_dir, Model.CHECKPOINT_FILENAME_PATTERN.format(step)
+        )
+        torch.save(self.state_dict(), path_to_checkpoint_file)
+        return path_to_checkpoint_file
+
+    def restore(self, path_to_checkpoint_file):
+        self.load_state_dict(
+            torch.load(path_to_checkpoint_file, map_location=torch.device("cpu"))
+        )
+        step = int(path_to_checkpoint_file.split("model-")[-1][:-4])
+        return step

requirements.txt

@@ -0,0 +1,4 @@
+gradio
+pillow
+torch
+torchvision