File size: 1,635 Bytes
6537db2 |
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 |
from typing import Dict, List, Any
import torch
import numpy as np
import torch.nn.functional as F
class EndpointHandler():
def __init__(self, path=""):
# load the optimized model
self.model = torch.load(path)
def __call__(self, data: Any) -> List[List[Dict[str, float]]]:
"""
Args:
data (:obj:):
includes the input data and the parameters for the inference.
Return:
A :obj:`list`:. The object returned should be a list of one list like [[{"label": 0.9939950108528137}]] containing :
- "label": A string representing what the label/class is. There can be multiple labels.
- "score": A score between 0 and 1 describing how confident the model is for this label/class.
"""
inputs = data.pop("inputs", data)
img = inputs["image"]
# Load the image
img = np.float32(img) / 255. # Load and normalize the image
# Convert to torch tensor and add batch dimension
img_tensor = torch.from_numpy(img).permute(2, 0, 1).unsqueeze(0)
# Padding if necessary (to make image dimensions multiples of 4)
b, c, h, w = img_tensor.shape
factor = 4 # Assuming factor is 4, based on the code context
H, W = ((h + factor) // factor) * factor, ((w + factor) // factor) * factor
padh = H - h if h % factor != 0 else 0
padw = W - w if w % factor != 0 else 0
img_tensor = F.pad(img_tensor, (0, padw, 0, padh), 'reflect')
restored = self.model(img_tensor)
# postprocess the prediction
return "OKAY" |