File size: 7,203 Bytes
233acb0
566678a
9188b68
2fb4fd2
9188b68
 
35037e4
2fb4fd2
 
 
 
 
 
 
 
 
35037e4
 
08ce2dc
566678a
2fb4fd2
35037e4
566678a
35037e4
2fb4fd2
 
 
35037e4
566678a
825b375
35037e4
9188b68
2fb4fd2
aa10251
9188b68
825b375
e1369ab
 
08ce2dc
 
9188b68
2fb4fd2
88b442b
 
35037e4
2fb4fd2
5ca6bfd
 
 
aa10251
566678a
aa10251
5ca6bfd
88b442b
 
08ce2dc
2fb4fd2
35037e4
88b442b
 
 
35037e4
 
08ce2dc
 
35037e4
aa10251
566678a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
aa10251
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
# handler.py  (repo root)
import io, base64, torch, open_clip
from PIL import Image
# optional: from open_clip import fuse_conv_bn_sequential   # if you want re‑param

class EndpointHandler:
    """
    MobileCLIP‑B ('datacompdr') zero‑shot classifier with per‑process
    text‑embedding cache.

    Expected client JSON:
      {
        "inputs": {
          "image": "<base64 PNG/JPEG>",
          "candidate_labels": ["a photo of a cat", ...]
        }
      }
    """

    def __init__(self, path=""):
        # Load the exact weights your local run uses
        self.model, _, self.preprocess = open_clip.create_model_and_transforms(
            "mobileclip_b", pretrained="datacompdr"
        )
        # Optional: fuse conv+bn for speed
        # self.model = fuse_conv_bn_sequential(self.model).eval()
        self.model.eval()

        self.tokenizer = open_clip.get_tokenizer("mobileclip_b")
        self.device = "cuda" if torch.cuda.is_available() else "cpu"
        self.model.to(self.device)

        self.cache: dict[str, torch.Tensor] = {}   # prompt → embedding

    def __call__(self, data):
        payload = data.get("inputs", data)
        img_b64 = payload["image"]
        labels  = payload.get("candidate_labels", [])
        if not labels:
            return {"error": "candidate_labels list is empty"}

        # Image → tensor
        img = Image.open(io.BytesIO(base64.b64decode(img_b64))).convert("RGB")
        img_t = self.preprocess(img).unsqueeze(0).to(self.device)

        # Text embeddings with cache
        new = [l for l in labels if l not in self.cache]
        if new:
            tok = self.tokenizer(new).to(self.device)
            with torch.no_grad():
                emb = self.model.encode_text(tok)
                emb = emb / emb.norm(dim=-1, keepdim=True)
            for l, e in zip(new, emb):
                self.cache[l] = e
        txt_t = torch.stack([self.cache[l] for l in labels])

        # Forward
        with torch.no_grad(), torch.cuda.amp.autocast():
            img_f = self.model.encode_image(img_t)
            img_f = img_f / img_f.norm(dim=-1, keepdim=True)
            probs = (100 * img_f @ txt_t.T).softmax(dim=-1)[0].tolist()

        return [
            {"label": l, "score": float(p)}
            for l, p in sorted(zip(labels, probs), key=lambda x: x[1], reverse=True)
        ]

# import io, base64, torch
# from PIL import Image
# import open_clip


# class EndpointHandler:
#     """
#     Zero‑shot classifier for MobileCLIP‑B (OpenCLIP) with a text‑embedding cache.

#     Client JSON:
#     {
#       "inputs": {
#         "image": "<base64 PNG/JPEG>",
#         "candidate_labels": ["cat", "dog", ...]
#       }
#     }
#     """

#     # ------------------------------------------------- #
#     #                 INITIALISATION                    #
#     # ------------------------------------------------- #
#     def __init__(self, path: str = ""):
#         weights = f"{path}/mobileclip_b.pt"

#         self.model, _, self.preprocess = open_clip.create_model_and_transforms(
#             "MobileCLIP-B", pretrained=weights
#         )
#         self.model.eval()

#         self.tokenizer = open_clip.get_tokenizer("MobileCLIP-B")
#         self.device = "cuda" if torch.cuda.is_available() else "cpu"
#         self.model.to(self.device)

#         # cache: {prompt -> 1×512 tensor on device}
#         self.label_cache: dict[str, torch.Tensor] = {}

#     # ------------------------------------------------- #
#     #                    INFERENCE                      #
#     # ------------------------------------------------- #
#     def __call__(self, data):
#         payload = data.get("inputs", data)

#         img_b64 = payload["image"]
#         labels  = payload.get("candidate_labels", [])
#         if not labels:
#             return {"error": "candidate_labels list is empty"}

#         # --- image ----
#         image = Image.open(io.BytesIO(base64.b64decode(img_b64))).convert("RGB")
#         img_tensor = self.preprocess(image).unsqueeze(0).to(self.device)

#         # --- text (with cache) ----
#         missing = [l for l in labels if l not in self.label_cache]
#         if missing:
#             tokens = self.tokenizer(missing).to(self.device)
#             with torch.no_grad():
#                 emb = self.model.encode_text(tokens)
#                 emb = emb / emb.norm(dim=-1, keepdim=True)
#             for l, e in zip(missing, emb):
#                 self.label_cache[l] = e
#         txt_feat = torch.stack([self.label_cache[l] for l in labels])

#         # --- forward & softmax ----
#         with torch.no_grad(), torch.cuda.amp.autocast():
#             img_feat = self.model.encode_image(img_tensor)
#             img_feat = img_feat / img_feat.norm(dim=-1, keepdim=True)
#             probs = (100 * img_feat @ txt_feat.T).softmax(dim=-1)[0].tolist()

#         # --- sorted output ----
#         return [
#             {"label": l, "score": float(p)}
#             for l, p in sorted(zip(labels, probs), key=lambda x: x[1], reverse=True)
#         ]

# # handler.py  (repo root)
# import io, base64, torch
# from PIL import Image
# import open_clip

# class EndpointHandler:
#     """
#     Zero‑shot classifier for MobileCLIP‑B (OpenCLIP).

#     Expected client JSON *to the endpoint*:
#     {
#       "inputs": {
#         "image": "<base64 PNG/JPEG>",
#         "candidate_labels": ["cat", "dog", ...]
#       }
#     }
#     """

#     def __init__(self, path: str = ""):
#         weights = f"{path}/mobileclip_b.pt"
#         self.model, _, self.preprocess = open_clip.create_model_and_transforms(
#             "MobileCLIP-B", pretrained=weights
#         )
#         self.model.eval()

#         self.tokenizer = open_clip.get_tokenizer("MobileCLIP-B")
#         self.device = "cuda" if torch.cuda.is_available() else "cpu"
#         self.model.to(self.device)

#     def __call__(self, data):
#         # ── unwrap Hugging Face's `inputs` envelope ───────────
#         payload = data.get("inputs", data)

#         img_b64 = payload["image"]
#         labels  = payload.get("candidate_labels", [])
#         if not labels:
#             return {"error": "candidate_labels list is empty"}

#         # Decode & preprocess image
#         image = Image.open(io.BytesIO(base64.b64decode(img_b64))).convert("RGB")
#         img_tensor = self.preprocess(image).unsqueeze(0).to(self.device)

#         # Tokenise labels
#         text_tokens = self.tokenizer(labels).to(self.device)

#         # Forward pass
#         with torch.no_grad(), torch.cuda.amp.autocast():
#             img_feat = self.model.encode_image(img_tensor)
#             txt_feat = self.model.encode_text(text_tokens)
#             img_feat = img_feat / img_feat.norm(dim=-1, keepdim=True)
#             txt_feat = txt_feat / txt_feat.norm(dim=-1, keepdim=True)
#             probs = (100 * img_feat @ txt_feat.T).softmax(dim=-1)[0].tolist()

#         # Sorted output
#         return [
#             {"label": l, "score": float(p)}
#             for l, p in sorted(zip(labels, probs), key=lambda x: x[1], reverse=True)
#         ]