File size: 9,559 Bytes
c9aa521
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
import os
import logging
import hashlib
import sys
import traceback
import copy
import tempfile

import cv2
import numpy as np
import torch
import torch.nn.functional as F
import gradio as gr
from PIL import Image, ImageFilter, ImageChops, ImageDraw
from huggingface_hub import hf_hub_download
import spaces

# --- IMPORT YOUR CUSTOM MODULES ---
from sam2.build_sam import build_sam2
from sam2.sam2_image_predictor import SAM2ImagePredictor
from plm_adapter_lora_with_image_input_only_text_positions import PLMLanguageAdapter

# ----------------- Configuration -----------------
SAM2_CONFIG = "sam2_hiera_l.yaml"
BASE_CKPT_NAME = "sam2_hiera_large.pt"

SQUARE_DIM = 1024
logging.basicConfig(level=logging.INFO)

# Refactored to store specific filenames per model choice
MODEL_CONFIGS = {
    "Stage 1": {
        "repo_id": "aadarsh99/ConvSeg-Stage1",
        "sam_filename": "fine_tuned_sam2_batched_100000.torch",
        "plm_filename": "fine_tuned_sam2_batched_plm_100000.torch"
    },
    "Stage 2 (grad-acc: 4)": {
        "repo_id": "aadarsh99/ConvSeg-Stage2",
        "sam_filename": "fine_tuned_sam2_batched_18000.torch",
        "plm_filename": "fine_tuned_sam2_batched_plm_18000.torch"
    },
    "Stage 2 (grad-acc: 8)": {
        "repo_id": "aadarsh99/ConvSeg-Stage2",
        "sam_filename": "fine_tuned_sam2_batched_18000.torch",
        "plm_filename": "fine_tuned_sam2_batched_plm_18000.torch"
    }
}

# Dynamically create cache keys based on config
MODEL_CACHE = {k: {"sam": None, "plm": None} for k in MODEL_CONFIGS.keys()}

# ----------------- Helper Functions -----------------
def download_if_needed(repo_id, filename):
    try:
        logging.info(f"Checking {filename} in {repo_id}...")
        return hf_hub_download(repo_id=repo_id, filename=filename)
    except Exception as e:
        raise FileNotFoundError(f"Could not find {filename} in {repo_id}. Error: {e}")

def stable_color(key: str):
    h = int(hashlib.sha256(str(key).encode("utf-8")).hexdigest(), 16)
    EDGE_COLORS_HEX = ["#3A86FF", "#FF006E", "#43AA8B", "#F3722C", "#8338EC", "#90BE6D"]
    colors = [tuple(int(c.lstrip("#")[i:i+2], 16) for i in (0, 2, 4)) for c in EDGE_COLORS_HEX]
    return colors[h % len(colors)]

def make_overlay(rgb: np.ndarray, mask: np.ndarray, key: str = "mask") -> Image.Image:
    # Convert base to RGBA
    base = Image.fromarray(rgb.astype(np.uint8)).convert("RGBA")
    mask_bool = mask > 0
    color = stable_color(key)
    
    # Create fill layer (Semi-transparent)
    fill_layer = Image.new("RGBA", base.size, color + (0,))
    fill_alpha = Image.fromarray((mask_bool.astype(np.uint8) * 140), "L")
    fill_layer.putalpha(fill_alpha)

    # Create stroke/edge layer
    m = Image.fromarray((mask_bool.astype(np.uint8) * 255), "L")
    edges = ImageChops.difference(m.filter(ImageFilter.MaxFilter(3)), m.filter(ImageFilter.MinFilter(3)))
    stroke_layer = Image.new("RGBA", base.size, color + (255,))
    stroke_layer.putalpha(edges)

    # Composite safely
    out = Image.alpha_composite(base, fill_layer)
    out = Image.alpha_composite(out, stroke_layer)
    
    return out.convert("RGB")

def ensure_models_loaded(stage_key):
    global MODEL_CACHE
    if MODEL_CACHE[stage_key]["sam"] is not None: 
        return
    
    config = MODEL_CONFIGS[stage_key]
    repo_id = config["repo_id"]
    
    logging.info(f"Loading {stage_key} models from {repo_id} into CPU RAM...")
    
    # SAM2
    # Base model is always the same
    base_path = download_if_needed(repo_id, BASE_CKPT_NAME)
    model = build_sam2(SAM2_CONFIG, base_path, device="cpu")
    
    # Load specific fine-tuned checkpoint
    final_path = download_if_needed(repo_id, config["sam_filename"])
    sd = torch.load(final_path, map_location="cpu")
    model.load_state_dict(sd.get("model", sd), strict=True)
    
    # PLM
    plm_path = download_if_needed(repo_id, config["plm_filename"])
    plm = PLMLanguageAdapter(
        model_name="Qwen/Qwen2.5-VL-3B-Instruct", 
        transformer_dim=model.sam_mask_decoder.transformer_dim, 
        n_sparse_tokens=0, use_dense_bias=True, use_lora=True, 
        lora_r=16, lora_alpha=32, lora_dropout=0.05, 
        dtype=torch.bfloat16, device="cpu"
    )
    plm_sd = torch.load(plm_path, map_location="cpu")
    plm.load_state_dict(plm_sd["plm"], strict=True)
    plm.eval()
    
    MODEL_CACHE[stage_key]["sam"], MODEL_CACHE[stage_key]["plm"] = model, plm

# ----------------- GPU Inference -----------------

@spaces.GPU(duration=120) 
def run_prediction(image_pil, text_prompt, threshold, stage_choice):
    if image_pil is None or not text_prompt:
        return None, None, None

    ensure_models_loaded(stage_choice)
    sam_model = MODEL_CACHE[stage_choice]["sam"]
    plm_model = MODEL_CACHE[stage_choice]["plm"]
    
    sam_model.to("cuda")
    plm_model.to("cuda")
    
    try:
        with torch.inference_mode():
            predictor = SAM2ImagePredictor(sam_model)
            rgb_orig = np.array(image_pil.convert("RGB"))
            H, W = rgb_orig.shape[:2]
            
            # Padding math
            scale = SQUARE_DIM / max(H, W)
            nw, nh = int(W * scale), int(H * scale)
            top, left = (SQUARE_DIM - nh) // 2, (SQUARE_DIM - nw) // 2

            # Resize & Pad
            rgb_sq = cv2.resize(rgb_orig, (nw, nh), interpolation=cv2.INTER_LINEAR)
            rgb_sq = cv2.copyMakeBorder(rgb_sq, top, SQUARE_DIM-nh-top, left, SQUARE_DIM-nw-left, cv2.BORDER_CONSTANT, value=0)
            
            predictor.set_image(rgb_sq)
            image_emb = predictor._features["image_embed"][-1].unsqueeze(0)
            hi = [lvl[-1].unsqueeze(0) for lvl in predictor._features["high_res_feats"]]

            # PLM adapter
            with tempfile.NamedTemporaryFile(suffix=".jpg") as tmp:
                image_pil.save(tmp.name)
                sp, dp = plm_model([text_prompt], image_emb.shape[2], image_emb.shape[3], [tmp.name])

            # SAM2 Decoding
            dec = sam_model.sam_mask_decoder
            dev, dtype = next(dec.parameters()).device, next(dec.parameters()).dtype
            
            low, scores, _, _ = dec(
                image_embeddings=image_emb.to(dev, dtype),
                image_pe=sam_model.sam_prompt_encoder.get_dense_pe().to(dev, dtype),
                sparse_prompt_embeddings=sp.to(dev, dtype),
                dense_prompt_embeddings=dp.to(dev, dtype),
                multimask_output=True, repeat_image=False,
                high_res_features=[h.to(dev, dtype) for h in hi]
            )

            # Postprocess to original dimensions
            logits = predictor._transforms.postprocess_masks(low, (SQUARE_DIM, SQUARE_DIM))
            best_idx = scores.argmax().item()
            logit_crop = logits[0, best_idx, top:top+nh, left:left+nw].unsqueeze(0).unsqueeze(0)
            logit_full = F.interpolate(logit_crop, size=(H, W), mode="bilinear", align_corners=False)[0, 0]
            
            prob = torch.sigmoid(logit_full).float().cpu().numpy()

        # Generate Heatmap
        heatmap_cv = cv2.applyColorMap((prob * 255).astype(np.uint8), cv2.COLORMAP_JET)
        heatmap_rgb = cv2.cvtColor(heatmap_cv, cv2.COLOR_BGR2RGB)
        
        # Initial Overlay
        mask = (prob > threshold).astype(np.uint8) * 255
        overlay = make_overlay(rgb_orig, mask, key=text_prompt)
        
        return overlay, Image.fromarray(heatmap_rgb), prob

    except Exception:
        traceback.print_exc()
        return None, None, None
    finally:
        sam_model.to("cpu")
        plm_model.to("cpu")
        torch.cuda.empty_cache()

def update_threshold_ui(image_pil, text_prompt, threshold, cached_prob):
    """Instant update using CPU only."""
    if image_pil is None or cached_prob is None:
        return None
    rgb_orig = np.array(image_pil.convert("RGB"))
    mask = (cached_prob > threshold).astype(np.uint8) * 255
    return make_overlay(rgb_orig, mask, key=text_prompt)

# ----------------- Gradio UI -----------------

with gr.Blocks(title="SAM2 + PLM Segmentation") as demo:
    prob_state = gr.State()
    
    gr.Markdown("# SAM2 + PLM Interactive Segmentation")
    gr.Markdown("Select a stage, enter a prompt, and run. Adjust the slider for **instant** mask updates.")
    
    with gr.Row():
        with gr.Column():
            input_image = gr.Image(type="pil", label="Input Image")
            text_prompt = gr.Textbox(label="Text Prompt", placeholder="e.g., 'the surgical forceps'")
            
            with gr.Row():
                stage_select = gr.Radio(
                    choices=list(MODEL_CONFIGS.keys()), 
                    value="Stage 2 (grad-acc: 8)", 
                    label="Model Stage"
                )
                threshold_slider = gr.Slider(0.0, 1.0, value=0.5, step=0.01, label="Threshold")
            
            run_btn = gr.Button("Run Inference", variant="primary")
        
        with gr.Column():
            out_overlay = gr.Image(label="Segmentation Overlay", type="pil")
            out_heatmap = gr.Image(label="Probability Heatmap", type="pil")

    # Full Pipeline
    run_btn.click(
        fn=run_prediction,
        inputs=[input_image, text_prompt, threshold_slider, stage_select],
        outputs=[out_overlay, out_heatmap, prob_state]
    )

    # Lightweight update on slider move
    threshold_slider.change(
        fn=update_threshold_ui,
        inputs=[input_image, text_prompt, threshold_slider, prob_state],
        outputs=[out_overlay]
    )

if __name__ == "__main__":
    demo.launch()