Spaces:
Runtime error
Runtime error
Upload 6 files
Browse files- .gitattributes +1 -1
- README.md +1 -2
- app.py +29 -23
- model.py +0 -2
- requirements.txt +2 -1
.gitattributes
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
*.pth filter=lfs diff=lfs merge=lfs -text
|
|
|
|
| 1 |
+
*.pth filter=lfs diff=lfs merge=lfs -text
|
README.md
CHANGED
|
@@ -10,7 +10,6 @@ app_file: app.py
|
|
| 10 |
pinned: false
|
| 11 |
---
|
| 12 |
|
| 13 |
-
|
| 14 |
# ResNet34 Corrosion Classifier — Hugging Face Space
|
| 15 |
|
| 16 |
Semplice Space Gradio che carica un modello ResNet34 e predice 9 classi di corrosione.
|
|
@@ -27,7 +26,7 @@ Semplice Space Gradio che carica un modello ResNet34 e predice 9 classi di corro
|
|
| 27 |
1. Crea una nuova Space su Hugging Face (Gradio + Python).
|
| 28 |
2. Carica questi file nella Space.
|
| 29 |
3. Aggiungi il tuo file di pesi `resnet34_best.pth` (usa Git LFS se > 50 MB).
|
| 30 |
-
4. (Opzionale) Se il file si chiama diversamente, imposta
|
| 31 |
nelle Settings della Space, oppure modifica `CKPT_PATH` in `app.py`.
|
| 32 |
5. Avvia la Space.
|
| 33 |
|
|
|
|
| 10 |
pinned: false
|
| 11 |
---
|
| 12 |
|
|
|
|
| 13 |
# ResNet34 Corrosion Classifier — Hugging Face Space
|
| 14 |
|
| 15 |
Semplice Space Gradio che carica un modello ResNet34 e predice 9 classi di corrosione.
|
|
|
|
| 26 |
1. Crea una nuova Space su Hugging Face (Gradio + Python).
|
| 27 |
2. Carica questi file nella Space.
|
| 28 |
3. Aggiungi il tuo file di pesi `resnet34_best.pth` (usa Git LFS se > 50 MB).
|
| 29 |
+
4. (Opzionale) Se il file si chiama diversamente, imposta la variabile d'ambiente `CKPT_PATH`
|
| 30 |
nelle Settings della Space, oppure modifica `CKPT_PATH` in `app.py`.
|
| 31 |
5. Avvia la Space.
|
| 32 |
|
app.py
CHANGED
|
@@ -12,14 +12,18 @@ from model import build_model, load_weights
|
|
| 12 |
TITLE = "ResNet34 Corrosion Classifier"
|
| 13 |
DESCRIPTION = """
|
| 14 |
Carica o scatta una foto. Il modello (ResNet34) restituisce la classe prevista e le probabilità.
|
| 15 |
-
Assicurati di
|
| 16 |
"""
|
| 17 |
|
| 18 |
-
# ====== Config ======
|
| 19 |
CKPT_PATH = os.environ.get("CKPT_PATH", "resnet34_best.pth")
|
| 20 |
CLASSES_PATH = os.environ.get("CLASSES_PATH", "classes.json")
|
| 21 |
-
DEVICE = "cpu"
|
| 22 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 23 |
with open(CLASSES_PATH, "r", encoding="utf-8") as f:
|
| 24 |
IDX2LABEL = json.load(f)
|
| 25 |
|
|
@@ -31,36 +35,38 @@ preprocess = transforms.Compose([
|
|
| 31 |
std=[0.229, 0.224, 0.225]),
|
| 32 |
])
|
| 33 |
|
| 34 |
-
# Lazy load del modello
|
| 35 |
_model = None
|
| 36 |
def get_model():
|
| 37 |
global _model
|
| 38 |
if _model is None:
|
| 39 |
-
model = build_model(num_classes=len(IDX2LABEL))
|
| 40 |
if not os.path.isfile(CKPT_PATH):
|
| 41 |
raise FileNotFoundError(
|
| 42 |
-
f
|
| 43 |
)
|
|
|
|
| 44 |
model = load_weights(model, CKPT_PATH, map_location=DEVICE)
|
| 45 |
_model = model
|
| 46 |
return _model
|
| 47 |
|
| 48 |
def predict(image: Image.Image, topk: int = 5):
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
|
|
|
|
|
|
|
|
|
|
| 64 |
|
| 65 |
with gr.Blocks(fill_height=True) as demo:
|
| 66 |
gr.Markdown(f"# {TITLE}")
|
|
@@ -69,10 +75,10 @@ with gr.Blocks(fill_height=True) as demo:
|
|
| 69 |
with gr.Row():
|
| 70 |
with gr.Column(scale=1):
|
| 71 |
img_in = gr.Image(type="pil", sources=["upload", "webcam"], label="Immagine")
|
| 72 |
-
topk = gr.Slider(1,
|
| 73 |
btn = gr.Button("Analizza immagine")
|
| 74 |
with gr.Column(scale=1):
|
| 75 |
-
lbl = gr.Label(label="Probabilità", num_top_classes=
|
| 76 |
txt = gr.Markdown()
|
| 77 |
|
| 78 |
btn.click(predict, inputs=[img_in, topk], outputs=[lbl, txt])
|
|
|
|
| 12 |
TITLE = "ResNet34 Corrosion Classifier"
|
| 13 |
DESCRIPTION = """
|
| 14 |
Carica o scatta una foto. Il modello (ResNet34) restituisce la classe prevista e le probabilità.
|
| 15 |
+
Assicurati di caricare il file dei pesi nella repo come `resnet34_best.pth` (o imposta la variabile di ambiente `CKPT_PATH`).
|
| 16 |
"""
|
| 17 |
|
|
|
|
| 18 |
CKPT_PATH = os.environ.get("CKPT_PATH", "resnet34_best.pth")
|
| 19 |
CLASSES_PATH = os.environ.get("CLASSES_PATH", "classes.json")
|
| 20 |
+
DEVICE = "cpu"
|
| 21 |
|
| 22 |
+
os.environ.setdefault("OMP_NUM_THREADS", "1")
|
| 23 |
+
os.environ.setdefault("MKL_NUM_THREADS", "1")
|
| 24 |
+
|
| 25 |
+
if not os.path.isfile(CLASSES_PATH):
|
| 26 |
+
raise FileNotFoundError(f"File classi non trovato: {CLASSES_PATH}")
|
| 27 |
with open(CLASSES_PATH, "r", encoding="utf-8") as f:
|
| 28 |
IDX2LABEL = json.load(f)
|
| 29 |
|
|
|
|
| 35 |
std=[0.229, 0.224, 0.225]),
|
| 36 |
])
|
| 37 |
|
|
|
|
| 38 |
_model = None
|
| 39 |
def get_model():
|
| 40 |
global _model
|
| 41 |
if _model is None:
|
|
|
|
| 42 |
if not os.path.isfile(CKPT_PATH):
|
| 43 |
raise FileNotFoundError(
|
| 44 |
+
f"Checkpoint non trovato: {CKPT_PATH}. Carica i pesi nella Space o imposta CKPT_PATH."
|
| 45 |
)
|
| 46 |
+
model = build_model(num_classes=len(IDX2LABEL))
|
| 47 |
model = load_weights(model, CKPT_PATH, map_location=DEVICE)
|
| 48 |
_model = model
|
| 49 |
return _model
|
| 50 |
|
| 51 |
def predict(image: Image.Image, topk: int = 5):
|
| 52 |
+
try:
|
| 53 |
+
if image is None:
|
| 54 |
+
return {}, "Nessuna immagine."
|
| 55 |
+
model = get_model()
|
| 56 |
+
model.eval()
|
| 57 |
+
with torch.no_grad():
|
| 58 |
+
img = image.convert("RGB")
|
| 59 |
+
tensor = preprocess(img).unsqueeze(0)
|
| 60 |
+
logits = model(tensor)
|
| 61 |
+
probs = torch.softmax(logits, dim=1).squeeze(0)
|
| 62 |
+
k = int(min(max(1, topk), probs.shape[0]))
|
| 63 |
+
values, indices = torch.topk(probs, k=k)
|
| 64 |
+
label_scores = {IDX2LABEL[i.item()]: float(v.item()) for v, i in zip(values, indices)}
|
| 65 |
+
pred_label = IDX2LABEL[int(torch.argmax(probs).item())]
|
| 66 |
+
msg = f"Predizione: **{pred_label}**"
|
| 67 |
+
return label_scores, msg
|
| 68 |
+
except Exception as e:
|
| 69 |
+
return {}, f"Errore durante l'inferenza: {e}"
|
| 70 |
|
| 71 |
with gr.Blocks(fill_height=True) as demo:
|
| 72 |
gr.Markdown(f"# {TITLE}")
|
|
|
|
| 75 |
with gr.Row():
|
| 76 |
with gr.Column(scale=1):
|
| 77 |
img_in = gr.Image(type="pil", sources=["upload", "webcam"], label="Immagine")
|
| 78 |
+
topk = gr.Slider(1, len(IDX2LABEL), value=5, step=1, label="Top-K")
|
| 79 |
btn = gr.Button("Analizza immagine")
|
| 80 |
with gr.Column(scale=1):
|
| 81 |
+
lbl = gr.Label(label="Probabilità", num_top_classes=len(IDX2LABEL))
|
| 82 |
txt = gr.Markdown()
|
| 83 |
|
| 84 |
btn.click(predict, inputs=[img_in, topk], outputs=[lbl, txt])
|
model.py
CHANGED
|
@@ -11,12 +11,10 @@ def build_model(num_classes: int) -> nn.Module:
|
|
| 11 |
|
| 12 |
def load_weights(model: nn.Module, ckpt_path: str, map_location="cpu") -> nn.Module:
|
| 13 |
state = torch.load(ckpt_path, map_location=map_location)
|
| 14 |
-
# Support both full state dicts and {'model': state_dict} formats
|
| 15 |
if isinstance(state, dict) and "state_dict" in state:
|
| 16 |
state = state["state_dict"]
|
| 17 |
if isinstance(state, dict) and "model" in state and isinstance(state["model"], dict):
|
| 18 |
state = state["model"]
|
| 19 |
-
# Strip possible 'module.' prefixes
|
| 20 |
new_state = {}
|
| 21 |
for k, v in state.items():
|
| 22 |
if k.startswith("module."):
|
|
|
|
| 11 |
|
| 12 |
def load_weights(model: nn.Module, ckpt_path: str, map_location="cpu") -> nn.Module:
|
| 13 |
state = torch.load(ckpt_path, map_location=map_location)
|
|
|
|
| 14 |
if isinstance(state, dict) and "state_dict" in state:
|
| 15 |
state = state["state_dict"]
|
| 16 |
if isinstance(state, dict) and "model" in state and isinstance(state["model"], dict):
|
| 17 |
state = state["model"]
|
|
|
|
| 18 |
new_state = {}
|
| 19 |
for k, v in state.items():
|
| 20 |
if k.startswith("module."):
|
requirements.txt
CHANGED
|
@@ -1,5 +1,6 @@
|
|
|
|
|
| 1 |
torch>=2.2.0
|
| 2 |
torchvision>=0.17.0
|
| 3 |
pillow>=10.3.0
|
| 4 |
numpy>=1.26.4
|
| 5 |
-
gradio==4.44.1
|
|
|
|
| 1 |
+
|
| 2 |
torch>=2.2.0
|
| 3 |
torchvision>=0.17.0
|
| 4 |
pillow>=10.3.0
|
| 5 |
numpy>=1.26.4
|
| 6 |
+
gradio==4.44.1
|