patent-wireframes / scripts /eval /condition_d_grounded.py
midah's picture
Reorganize: scripts/eval/condition_d_grounded.py
35e4d12 verified
"""Condition D: Vision + grounded annotation — image AND masked text description.
Task 1 — Viewpoint Identification:
Given a representation of one patent figure, name its viewpoint type.
Ground truth: parsed from drawing_description text.
Condition D setup:
The VLM sees BOTH:
1. The image of the target figure (visual signal)
2. The drawing description with the target viewpoint MASKED (FIM text context)
e.g. "FIG. 1 is perspective; FIG. 3 is rear elevational; FIG. 2 is ___ view"
3. Numeric features (ink density, edge transitions, aspect ratio)
This is the CadVLM setup: image + structured text together.
Hypothesis: Condition D > Condition A (image only) because the text context
scaffolds convention knowledge onto the visual signal.
Comparison targets:
Condition A (vision only): 23.3% overall, 5.3% front elev
Condition C (text FIM, no image): see condition_c_results.json
Condition D (vision + text FIM): this script
Usage:
python scripts/eval/condition_d_grounded.py \
--enriched data/enriched/enriched_2022.parquet \
--images /tmp/patent_sample/2022 \
--n 18 \
--out results/condition_d_results.json
"""
import argparse
import json
import re
import time
from collections import defaultdict
from pathlib import Path
import pandas as pd
import sys
sys.path.insert(0, str(Path(__file__).parent))
from provider import chat, encode_image, get_client
# ── shared helpers (same as condition_c and track_a) ─────────────────────────
def parse_viewpoint(desc: str, fig_num: int) -> str:
pat = re.compile(
rf"FIG\.\s*{fig_num + 1}\s+is\s+(?:a\s+|an\s+)?(.{{5,80}}?)\s*(?:view|thereof|;|\n|$)",
re.IGNORECASE,
)
m = pat.search(desc or "")
return m.group(1).strip().lower() if m else ""
def mask_viewpoint(desc: str, fig_num: int) -> tuple[str, str]:
pat = re.compile(
rf"(FIG\.\s*{fig_num + 1}\s+is\s+(?:a\s+|an\s+)?)(.{{5,80}}?)(\s*(?:view|thereof|;|\n|$))",
re.IGNORECASE,
)
vp = parse_viewpoint(desc, fig_num)
masked = pat.sub(r"\1___\3", desc or "", count=1)
return masked, vp
def viewpoint_match(predicted: str, ground_truth: str) -> bool:
DIRECTIONS = {"front","rear","back","left","right","top","bottom","side",
"perspective","plan","elevation","elevational","isometric",
"oblique","reference","detail","enlarged","cross","section"}
pred_w = set(re.findall(r"\w+", predicted.lower())) & DIRECTIONS
gt_w = set(re.findall(r"\w+", ground_truth.lower())) & DIRECTIONS
if not gt_w: return False
return len(pred_w & gt_w) / len(gt_w) >= 0.5
VP_BUCKETS = [
("perspective", lambda v: "perspective" in v),
("front_elev", lambda v: "front elev" in v),
("rear_elev", lambda v: "rear elev" in v or "rear elevation" in v),
("side_elev", lambda v: "side elev" in v or "side elevation" in v),
("top_plan", lambda v: "top plan" in v or v == "top view"),
("bottom_plan", lambda v: "bottom plan" in v or "bottom" in v),
("other", lambda v: True),
]
def bucket(vp: str) -> str:
for name, fn in VP_BUCKETS:
if fn(vp): return name
return "other"
def find_image(images_dir: Path, image_filename: str) -> Path | None:
parts = image_filename.split("-D0")
if len(parts) < 2: return None
p = images_dir / parts[0] / image_filename
return p if p.exists() else None
# ── Condition D prompt: image + masked text + numeric features ────────────────
TASK1_D_PROMPT = """\
You are an engineer reading a US design patent.
Patent title: {title}
The patent has {n_figs} figures. Here is the drawing description with one \
figure's viewpoint type replaced by ___:
{masked_desc}
{numeric_context}
The image above shows FIG. {fig_num} of this patent.
Using both the image and the context of what other figures show, what type \
of view is FIG. {fig_num}?
Give the viewpoint label only (2-5 words).
Examples: "front elevational view", "perspective view", "top plan view"
"""
def run(enriched_path: str, images_dir: str, n: int, out_path: str, seed: int = 42):
import random
client = get_client()
images = Path(images_dir)
df = pd.read_parquet(enriched_path)
df["vp"] = df.apply(
lambda r: parse_viewpoint(r.get("drawing_description", ""), r["figure_number"]),
axis=1,
)
df["img_path"] = df["image_filename"].apply(lambda fn: find_image(images, fn))
has_persp = df.groupby("patent_id")["vp"].apply(lambda x: x.str.contains("perspective").any())
has_front = df.groupby("patent_id")["vp"].apply(
lambda x: (x.str.contains("front elev") | x.isin(["front elevation","front plan"])).any()
)
has_img = df.groupby("patent_id")["img_path"].apply(lambda x: x.notna().any())
eligible = has_persp.index[has_persp & has_front & has_img].tolist()
rng = random.Random(seed)
rng.shuffle(eligible)
eval_pids = eligible[:n]
print(f"Eligible patents with images: {len(eligible)}, evaluating: {len(eval_pids)}")
results, by_vp = [], defaultdict(lambda: [0, 0])
total_correct = total_n = 0
for pid in eval_pids:
group = df[df["patent_id"] == pid].sort_values("figure_number")
patent_results = []
for _, row in group.iterrows():
img_path = row.get("img_path")
if img_path is None:
continue
desc = str(row.get("drawing_description") or "")
fig_num = int(row.get("figure_number", 0))
masked_desc, ground_truth = mask_viewpoint(desc, fig_num)
if not ground_truth:
continue
# Numeric context
numeric_parts = []
for col, label in [("ink_frac","ink density"),("edge_transitions","edges"),
("img_aspect","aspect ratio")]:
val = row.get(col)
if val is not None and not pd.isna(val):
if col == "ink_frac": numeric_parts.append(f" {label}: {float(val):.2%}")
elif col == "edge_transitions": numeric_parts.append(f" {label}: {int(val):,}")
else: numeric_parts.append(f" {label}: {float(val):.2f}")
numeric_context = ("Image statistics:\n" + "\n".join(numeric_parts)
if numeric_parts else "")
text_prompt = TASK1_D_PROMPT.format(
title=str(row.get("patent_title", "")),
n_figs=int(row.get("n_figures_in_patent", 0)),
masked_desc=masked_desc.strip(),
numeric_context=numeric_context,
fig_num=fig_num + 1,
)
# ── TASK 1 (Condition D): image + grounded text ───────────────────
# Input: image of the figure (visual) + masked drawing description
# (FIM text context) + numeric features
# Output: predicted viewpoint type
# Model uses BOTH the visual appearance AND the set context
b64, media = encode_image(Path(img_path))
from provider import multi_image_message
msgs = multi_image_message(
images=[(b64, media)],
text_after=text_prompt,
)
predicted = chat(client, msgs, max_tokens=30).lower().strip()
correct = viewpoint_match(predicted, ground_truth)
vp_label = bucket(ground_truth)
by_vp[vp_label][1] += 1
if correct: by_vp[vp_label][0] += 1
total_correct += int(correct)
total_n += 1
patent_results.append({
"fig": fig_num, "ground_truth": ground_truth,
"predicted": predicted, "correct": correct, "vp_bucket": vp_label,
})
time.sleep(0.3)
if patent_results:
n_c = sum(r["correct"] for r in patent_results)
results.append({"patent_id": pid,
"patent_title": str(group["patent_title"].iloc[0]),
"figures": patent_results})
print(f"[{pid}] {str(group['patent_title'].iloc[0])[:45]} — "
f"{n_c}/{len(patent_results)} correct")
print()
print("=" * 60)
print("CONDITION D — VISION + GROUNDED ANNOTATION RESULTS")
print("Input: image + masked drawing description (FIM) + numeric features")
print("=" * 60)
print(f"Overall: {total_correct}/{total_n} = {total_correct/max(total_n,1):.1%}")
print()
for label, (c, t) in sorted(by_vp.items(), key=lambda x: -x[1][1]):
if t == 0: continue
print(f" {label:<14} {c:>3}/{t:<3} {c/t:>5.1%}")
output = {
"condition": "D_vision_plus_text",
"task": "Task 1 — Viewpoint Identification",
"input": "image + drawing_description FIM + numeric features",
"summary": {
"overall_acc": total_correct / max(total_n, 1),
"n": total_n,
"by_viewpoint": {k: {"correct": v[0], "total": v[1], "acc": v[0]/v[1] if v[1] else 0}
for k, v in by_vp.items()},
},
"results": results,
}
Path(out_path).parent.mkdir(parents=True, exist_ok=True)
with open(out_path, "w") as f:
json.dump(output, f, indent=2)
print(f"\nResults → {out_path}")
def main():
parser = argparse.ArgumentParser()
parser.add_argument("--enriched", default="data/enriched/enriched_2022.parquet")
parser.add_argument("--images", default="/tmp/patent_sample/2022")
parser.add_argument("--n", type=int, default=18)
parser.add_argument("--out", default="results/condition_d_results.json")
args = parser.parse_args()
run(args.enriched, args.images, args.n, args.out)
if __name__ == "__main__":
main()