progress-photo-analyzer

What this skill does

# Progress Photo Analyzer

## Business Case

### Problem Statement
Site photos are underutilized for progress tracking:
- Manual review is time-consuming
- Subjective progress assessment
- No systematic comparison to plans
- Safety issues may be missed

### Solution
AI-powered photo analysis system that extracts progress information, detects safety concerns, and compares site conditions to BIM models.

### Business Value
- **Automation** - Reduce manual photo review
- **Accuracy** - Objective progress measurement
- **Safety** - Automatic hazard detection
- **Documentation** - Structured photo records

## Technical Implementation

```python
import pandas as pd
from datetime import datetime, date
from typing import Dict, Any, List, Optional, Tuple
from dataclasses import dataclass, field
from enum import Enum
from pathlib import Path
import base64


class PhotoType(Enum):
    """Types of construction photos."""
    PROGRESS = "progress"
    SAFETY = "safety"
    QUALITY = "quality"
    GENERAL = "general"
    DELIVERY = "delivery"


class AnalysisStatus(Enum):
    """Analysis status."""
    PENDING = "pending"
    ANALYZING = "analyzing"
    COMPLETED = "completed"
    FAILED = "failed"


class SafetyIssue(Enum):
    """Detected safety issues."""
    MISSING_PPE = "missing_ppe"
    FALL_HAZARD = "fall_hazard"
    HOUSEKEEPING = "housekeeping"
    SCAFFOLDING = "scaffolding"
    ELECTRICAL = "electrical"
    EXCAVATION = "excavation"
    NONE = "none"


class WorkActivity(Enum):
    """Detected work activities."""
    EXCAVATION = "excavation"
    FOUNDATION = "foundation"
    CONCRETE_POUR = "concrete_pour"
    STEEL_ERECTION = "steel_erection"
    FRAMING = "framing"
    ROOFING = "roofing"
    MEP_ROUGH = "mep_rough"
    DRYWALL = "drywall"
    FINISHES = "finishes"
    EXTERIOR = "exterior"
    UNKNOWN = "unknown"


@dataclass
class PhotoMetadata:
    """Photo metadata."""
    photo_id: str
    filename: str
    capture_date: datetime
    location: str
    level: str
    zone: str
    photo_type: PhotoType
    photographer: str = ""
    gps_coordinates: Optional[Tuple[float, float]] = None
    file_path: str = ""


@dataclass
class ProgressDetection:
    """Detected progress information."""
    work_activity: WorkActivity
    confidence: float
    description: str
    completion_estimate: float  # 0-100%
    elements_visible: List[str] = field(default_factory=list)


@dataclass
class SafetyDetection:
    """Detected safety information."""
    issue_type: SafetyIssue
    confidence: float
    description: str
    severity: str  # low, medium, high
    location_in_image: Optional[Tuple[int, int, int, int]] = None  # bounding box


@dataclass
class PhotoAnalysisResult:
    """Complete photo analysis result."""
    photo_id: str
    metadata: PhotoMetadata
    analysis_date: datetime
    status: AnalysisStatus
    progress_detections: List[ProgressDetection]
    safety_detections: List[SafetyDetection]
    weather_conditions: str
    worker_count: int
    equipment_visible: List[str]
    quality_issues: List[str]
    notes: str = ""
    bim_comparison: Optional[Dict[str, Any]] = None


class ProgressPhotoAnalyzer:
    """Analyze construction site photos."""

    def __init__(self, project_name: str):
        self.project_name = project_name
        self.photos: Dict[str, PhotoMetadata] = {}
        self.results: Dict[str, PhotoAnalysisResult] = {}
        self._photo_counter = 0

    def register_photo(self,
                      filename: str,
                      capture_date: datetime,
                      location: str,
                      level: str = "",
                      zone: str = "",
                      photo_type: PhotoType = PhotoType.PROGRESS,
                      photographer: str = "",
                      file_path: str = "") -> PhotoMetadata:
        """Register a photo for analysis."""
        self._photo_counter += 1
        photo_id = f"PH-{self._photo_counter:05d}"

        metadata = PhotoMetadata(
            photo_id=photo_id,
            filename=filename,
            capture_date=capture_date,
            location=location,
            level=level,
            zone=zone,
            photo_type=photo_type,
            photographer=photographer,
            file_path=file_path
        )

        self.photos[photo_id] = metadata
        return metadata

    def analyze_photo(self, photo_id: str,
                     image_data: bytes = None) -> PhotoAnalysisResult:
        """Analyze a registered photo."""
        if photo_id not in self.photos:
            raise ValueError(f"Photo {photo_id} not registered")

        metadata = self.photos[photo_id]

        # Perform analysis (simulated - would use CV/AI models)
        progress_detections = self._detect_progress(metadata, image_data)
        safety_detections = self._detect_safety(metadata, image_data)
        weather = self._detect_weather(metadata, image_data)
        worker_count = self._count_workers(image_data)
        equipment = self._detect_equipment(image_data)

        result = PhotoAnalysisResult(
            photo_id=photo_id,
            metadata=metadata,
            analysis_date=datetime.now(),
            status=AnalysisStatus.COMPLETED,
            progress_detections=progress_detections,
            safety_detections=safety_detections,
            weather_conditions=weather,
            worker_count=worker_count,
            equipment_visible=equipment,
            quality_issues=[]
        )

        self.results[photo_id] = result
        return result

    def _detect_progress(self, metadata: PhotoMetadata,
                        image_data: bytes = None) -> List[ProgressDetection]:
        """Detect work progress in photo."""
        # Simulated detection based on metadata
        detections = []

        # In real implementation, this would use computer vision
        location_lower = metadata.location.lower()

        if 'foundation' in location_lower or 'basement' in location_lower:
            detections.append(ProgressDetection(
                work_activity=WorkActivity.FOUNDATION,
                confidence=0.85,
                description="Foundation work visible",
                completion_estimate=60.0
            ))
        elif 'steel' in location_lower or 'structure' in location_lower:
            detections.append(ProgressDetection(
                work_activity=WorkActivity.STEEL_ERECTION,
                confidence=0.90,
                description="Structural steel installation",
                completion_estimate=45.0
            ))
        elif 'roof' in location_lower:
            detections.append(ProgressDetection(
                work_activity=WorkActivity.ROOFING,
                confidence=0.80,
                description="Roofing work in progress",
                completion_estimate=30.0
            ))
        else:
            detections.append(ProgressDetection(
                work_activity=WorkActivity.UNKNOWN,
                confidence=0.50,
                description="General construction activity",
                completion_estimate=0.0
            ))

        return detections

    def _detect_safety(self, metadata: PhotoMetadata,
                      image_data: bytes = None) -> List[SafetyDetection]:
        """Detect safety issues in photo."""
        # Simulated detection - real implementation would use AI models
        detections = []

        # In production, this would analyze the actual image
        if metadata.photo_type == PhotoType.SAFETY:
            # Return empty for demonstration
            pass

        return detections

    def _detect_weather(self, metadata: PhotoMetadata,
                       image_data: bytes = None) -> str:
        """Detect weather conditions from photo."""
        # Simulated - would use image analysis
        return "clear"

    def _count_workers(self, image_data: bytes = None) -> int:
        """Count workers visible in photo."""
        # Simulated