coralnet-toolbox 0.0.74__tar.gz → 0.0.76__tar.gz

{coralnet_toolbox-0.0.74 → coralnet_toolbox-0.0.76}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: coralnet-toolbox
-Version: 0.0.74
+Version: 0.0.76
 Summary: Tools for annotating and developing ML models for benthic imagery
 Author-email: Jordan Pierce <jordan.pierce@noaa.gov>
 License: MIT License
@@ -16,7 +16,7 @@ License-File: LICENSE.txt
 Requires-Dist: PyQt5>=5.15.11
 Requires-Dist: pyqtdarktheme
 Requires-Dist: pyqtgraph
-Requires-Dist: ultralytics>=8.3.152
+Requires-Dist: ultralytics>=8.3.191
 Requires-Dist: lap>=0.5.12
 Requires-Dist: open-clip-torch>=2.20.0
 Requires-Dist: supervision>=0.24.0
@@ -27,7 +27,8 @@ Requires-Dist: pycocotools
 Requires-Dist: ujson
 Requires-Dist: timm==0.9.2
 Requires-Dist: autodistill
-Requires-Dist: transformers>=4.5.0
+Requires-Dist: transformers>=4.56.0
+Requires-Dist: hf_xet
 Requires-Dist: x-segment-anything>=0.0.8
 Requires-Dist: yolo-tiling>=0.0.19
 Requires-Dist: roboflow
@@ -38,6 +39,8 @@ Requires-Dist: beautifulsoup4>=4.12.2
 Requires-Dist: webdriver_manager
 Requires-Dist: dill
 Requires-Dist: seaborn
+Requires-Dist: GPUtil
+Requires-Dist: psutil
 Provides-Extra: all
 Requires-Dist: coralnet-toolbox[extra]; extra == "all"
 Dynamic: license-file
@@ -113,6 +116,7 @@ For a complete installation guide (including CUDA setup), see the [Installation
 | **Overview** | Get the big picture | [📋 Read More](https://jordan-pierce.github.io/CoralNet-Toolbox/overview) |
 | **Installation** | Detailed setup instructions | [⚙️ Setup Guide](https://jordan-pierce.github.io/CoralNet-Toolbox/installation) |
 | **Usage** | Learn the tools | [🛠️ User Manual](https://jordan-pierce.github.io/CoralNet-Toolbox/usage) |
+| **Hot Keys** | Keyboard shortcuts | [⌨️ Shortcuts](https://jordan-pierce.github.io/CoralNet-Toolbox/hot-keys) |
 | **Classification** | Community tutorial | [🧠 AI Tutorial](https://jordan-pierce.github.io/CoralNet-Toolbox/classify) |
 
 </div>
@@ -178,7 +182,7 @@ The toolbox integrates state-of-the-art models for efficient annotation workflow
 | **Framework** | **Models** | **Capability** |
 |:---:|:---:|:---:|
 | **YOLOE** | See Anything | Visual prompt detection |
-| **AutoDistill** | Grounding DINO • OWLViT • OmDetTurbo | Zero-shot detection |
+| **Transformers** | Grounding DINO • OWLViT • OmDetTurbo | Zero-shot detection |
 
 </div>
 
@@ -276,12 +280,9 @@ uv pip install coralnet-toolbox
 ### 🚀 **Step 3: GPU Acceleration (Optional)**
 For CUDA-enabled systems:
 ```bash
-# Example for CUDA 11.8
-conda install nvidia/label/cuda-11.8.0::cuda-nvcc -y
-conda install nvidia/label/cuda-11.8.0::cuda-toolkit -y
-
+# Example for CUDA 12.9
 # Install PyTorch with CUDA support
-uv pip install torch torchvision --index-url https://download.pytorch.org/whl/cu118 --upgrade
+uv pip install torch torchvision --index-url https://download.pytorch.org/whl/cu129 --upgrade
 ```
 
 ### 🏃‍♂️ **Step 4: Launch**
@@ -308,7 +309,9 @@ uv pip install -U coralnet-toolbox==[latest_version]
 ### 🏗️ **Repository Structure**
 
 <div align="center">
-![Visualization of the codebase](./diagram.svg)
+  <a href="https://raw.githubusercontent.com/Jordan-Pierce/CoralNet-Toolbox/refs/heads/main/diagram.svg">
+    <img src="https://raw.githubusercontent.com/Jordan-Pierce/CoralNet-Toolbox/refs/heads/main/diagram.svg" alt="Visualization of the codebase" width="80%">
+  </a>
 </div>
 
 ---

{coralnet_toolbox-0.0.74 → coralnet_toolbox-0.0.76}/README.md

@@ -69,6 +69,7 @@ For a complete installation guide (including CUDA setup), see the [Installation
 | **Overview** | Get the big picture | [📋 Read More](https://jordan-pierce.github.io/CoralNet-Toolbox/overview) |
 | **Installation** | Detailed setup instructions | [⚙️ Setup Guide](https://jordan-pierce.github.io/CoralNet-Toolbox/installation) |
 | **Usage** | Learn the tools | [🛠️ User Manual](https://jordan-pierce.github.io/CoralNet-Toolbox/usage) |
+| **Hot Keys** | Keyboard shortcuts | [⌨️ Shortcuts](https://jordan-pierce.github.io/CoralNet-Toolbox/hot-keys) |
 | **Classification** | Community tutorial | [🧠 AI Tutorial](https://jordan-pierce.github.io/CoralNet-Toolbox/classify) |
 
 </div>
@@ -134,7 +135,7 @@ The toolbox integrates state-of-the-art models for efficient annotation workflow
 | **Framework** | **Models** | **Capability** |
 |:---:|:---:|:---:|
 | **YOLOE** | See Anything | Visual prompt detection |
-| **AutoDistill** | Grounding DINO • OWLViT • OmDetTurbo | Zero-shot detection |
+| **Transformers** | Grounding DINO • OWLViT • OmDetTurbo | Zero-shot detection |
 
 </div>
 
@@ -232,12 +233,9 @@ uv pip install coralnet-toolbox
 ### 🚀 **Step 3: GPU Acceleration (Optional)**
 For CUDA-enabled systems:
 ```bash
-# Example for CUDA 11.8
-conda install nvidia/label/cuda-11.8.0::cuda-nvcc -y
-conda install nvidia/label/cuda-11.8.0::cuda-toolkit -y
-
+# Example for CUDA 12.9
 # Install PyTorch with CUDA support
-uv pip install torch torchvision --index-url https://download.pytorch.org/whl/cu118 --upgrade
+uv pip install torch torchvision --index-url https://download.pytorch.org/whl/cu129 --upgrade
 ```
 
 ### 🏃‍♂️ **Step 4: Launch**
@@ -264,7 +262,9 @@ uv pip install -U coralnet-toolbox==[latest_version]
 ### 🏗️ **Repository Structure**
 
 <div align="center">
-![Visualization of the codebase](./diagram.svg)
+  <a href="https://raw.githubusercontent.com/Jordan-Pierce/CoralNet-Toolbox/refs/heads/main/diagram.svg">
+    <img src="https://raw.githubusercontent.com/Jordan-Pierce/CoralNet-Toolbox/refs/heads/main/diagram.svg" alt="Visualization of the codebase" width="80%">
+  </a>
 </div>
 
 ---

{coralnet_toolbox-0.0.74 → coralnet_toolbox-0.0.76}/coralnet_toolbox/Annotations/QtPolygonAnnotation.py

@@ -306,20 +306,30 @@ class PolygonAnnotation(Annotation):
         self.set_cropped_bbox()
         # Get the bounding box of the polygon
         min_x, min_y, max_x, max_y = self.cropped_bbox
-
+        
+        # Ensure min/max values are correctly ordered
+        min_x, max_x = min(min_x, max_x), max(min_x, max_x)
+        min_y, max_y = min(min_y, max_y), max(min_y, max_y)
+        
+        # Clamp values to image bounds
+        min_x = max(0, min(rasterio_src.width - 1, min_x))
+        min_y = max(0, min(rasterio_src.height - 1, min_y))
+        max_x = max(min_x + 1, min(rasterio_src.width, max_x))
+        max_y = max(min_y + 1, min(rasterio_src.height, max_y))
+    
         # Calculate the window for rasterio
         window = Window(
-            col_off=max(0, int(min_x)),
-            row_off=max(0, int(min_y)),
-            width=min(rasterio_src.width - int(min_x), int(max_x - min_x)),
-            height=min(rasterio_src.height - int(min_y), int(max_y - min_y))
+            col_off=int(min_x),
+            row_off=int(min_y),
+            width=int(max_x - min_x),
+            height=int(max_y - min_y)
         )
-
+    
         # Convert rasterio to QImage
         q_image = rasterio_to_cropped_image(self.rasterio_src, window)
         # Convert QImage to QPixmap
         self.cropped_image = QPixmap.fromImage(q_image)
-
+    
         self.annotationUpdated.emit(self)  # Notify update
 
     def create_graphics_item(self, scene: QGraphicsScene):
@@ -423,7 +433,7 @@ class PolygonAnnotation(Annotation):
         self.update_graphics_item()
         self.annotationUpdated.emit(self)  # Notify update
 
-    def update_annotation_size(self, delta: float):
+    def update_annotation_size(self, scale_factor: float):
         """
         Grow/shrink the polygon and its holes by scaling vertices radially from the centroid.
         """
@@ -433,18 +443,53 @@ class PolygonAnnotation(Annotation):
             return
 
         # 1. Use the true geometric centroid as the pivot for scaling.
-        # This is correctly calculated by the new set_centroid() method.
         centroid_x = self.center_xy.x()
         centroid_y = self.center_xy.y()
 
-        # 2. Determine the scale factor (this logic remains the same).
+        # 2. Determine the scale factor.
         step = 0.01  # Adjust for finer or coarser changes
-        if delta > 1.0:
+        if scale_factor > 1.0:
             scale = 1.0 + step
-        elif delta < 1.0:
+        elif scale_factor < 1.0:
             scale = 1.0 - step
         else:
             scale = 1.0
+        
+        # Check for image boundaries before scaling
+        if hasattr(self, 'rasterio_src'):
+            # Before scaling, check if any point would go beyond image boundaries
+            img_width = self.rasterio_src.width
+            img_height = self.rasterio_src.height
+            padding = 10  # pixels of padding
+            
+            # Check outer boundary points
+            for p in self.points:
+                dx = p.x() - centroid_x
+                dy = p.y() - centroid_y
+                new_x = centroid_x + dx * scale
+                new_y = centroid_y + dy * scale
+                
+                if new_x < -padding or \
+                   new_y < -padding or \
+                   new_x > img_width + padding or \
+                   new_y > img_height + padding:
+                    # Point would be out of bounds, don't scale
+                    return
+            
+            # Check holes points too
+            for hole in self.holes:
+                for p in hole:
+                    dx = p.x() - centroid_x
+                    dy = p.y() - centroid_y
+                    new_x = centroid_x + dx * scale
+                    new_y = centroid_y + dy * scale
+                    
+                    if new_x < -padding or \
+                       new_y < -padding or \
+                       new_x > img_width + padding or \
+                       new_y > img_height + padding:
+                        # Point would be out of bounds, don't scale
+                        return
 
         # 3. Scale the outer boundary points.
         new_points = []

{coralnet_toolbox-0.0.74 → coralnet_toolbox-0.0.76}/coralnet_toolbox/Annotations/QtRectangleAnnotation.py

@@ -213,23 +213,29 @@ class RectangleAnnotation(Annotation):
         # Get the bounding box of the rectangle
         min_x, min_y, max_x, max_y = self.cropped_bbox
         
-        # Ensure min/max values are correctly ordered to avoid negative width/height
+        # Ensure min/max values are correctly ordered
         min_x, max_x = min(min_x, max_x), max(min_x, max_x)
         min_y, max_y = min(min_y, max_y), max(min_y, max_y)
-
+        
+        # Clamp values to image bounds
+        min_x = max(0, min(rasterio_src.width - 1, min_x))
+        min_y = max(0, min(rasterio_src.height - 1, min_y))
+        max_x = max(min_x + 1, min(rasterio_src.width, max_x))
+        max_y = max(min_y + 1, min(rasterio_src.height, max_y))
+        
         # Calculate the window for rasterio
         window = Window(
-            col_off=max(0, int(min_x)),
-            row_off=max(0, int(min_y)),
-            width=min(rasterio_src.width - int(min_x), int(max_x - min_x)),
-            height=min(rasterio_src.height - int(min_y), int(max_y - min_y))
+            col_off=int(min_x),
+            row_off=int(min_y),
+            width=int(max_x - min_x),
+            height=int(max_y - min_y)
         )
-
+        
         # Convert rasterio to QImage
         q_image = rasterio_to_cropped_image(self.rasterio_src, window)
         # Convert QImage to QPixmap
         self.cropped_image = QPixmap.fromImage(q_image)
-
+        
         self.annotationUpdated.emit(self)  # Notify update
 
     def create_graphics_item(self, scene: QGraphicsScene):
@@ -282,13 +288,37 @@ class RectangleAnnotation(Annotation):
         # Clear the machine confidence
         self.update_user_confidence(self.label)
 
-        # Update the size, graphic
-        width = (self.bottom_right.x() - self.top_left.x()) * scale_factor
-        height = (self.bottom_right.y() - self.top_left.y()) * scale_factor
-        self.top_left = QPointF(self.center_xy.x() - width / 2, self.center_xy.y() - height / 2)
-        self.bottom_right = QPointF(self.center_xy.x() + width / 2, self.center_xy.y() + height / 2)
+        # Calculate new dimensions
+        current_width = self.bottom_right.x() - self.top_left.x()
+        current_height = self.bottom_right.y() - self.top_left.y()
+        new_width = current_width * scale_factor
+        new_height = current_height * scale_factor
+        
+        # Calculate tentative new corners
+        new_left = self.center_xy.x() - new_width / 2
+        new_top = self.center_xy.y() - new_height / 2
+        new_right = self.center_xy.x() + new_width / 2
+        new_bottom = self.center_xy.y() + new_height / 2
+        
+        # Check if we're within valid image bounds (or use a reasonable padding)
+        # Assuming we have access to the image dimensions
+        if hasattr(self, 'rasterio_src'):
+            img_width = self.rasterio_src.width
+            img_height = self.rasterio_src.height
+            
+            # Apply reasonable bounds with some margin
+            padding = 10  # pixels
+            if new_left < -padding or new_top < -padding or new_right > img_width + padding or new_bottom > img_height + padding:
+                # Don't resize beyond these limits
+                return
+        
+        # Update the rectangle coordinates
+        self.top_left = QPointF(new_left, new_top)
+        self.bottom_right = QPointF(new_right, new_bottom)
+        
+        # Update graphics and notify
         self.update_graphics_item()
-        self.annotationUpdated.emit(self)  # Notify update
+        self.annotationUpdated.emit(self)
 
     def resize(self, handle: str, new_pos: QPointF):
         """Resize the annotation based on the handle and new position."""

{coralnet_toolbox-0.0.74 → coralnet_toolbox-0.0.76}/coralnet_toolbox/Explorer/QtDataItem.py

@@ -121,24 +121,45 @@ class AnnotationImageWidget(QWidget):
         self.setToolTip(self.data_item.get_tooltip_text())
 
     def recalculate_aspect_ratio(self):
-        """Calculate aspect ratio from annotation geometry without loading image."""
+        """Calculate and store the annotation's aspect ratio."""
+        annotation = self.data_item.annotation
+        
+        # Try to use the cropped_bbox attribute first
+        if hasattr(annotation, 'cropped_bbox'):
+            min_x, min_y, max_x, max_y = annotation.cropped_bbox
+            width = max_x - min_x
+            height = max_y - min_y
+            
+            if height > 0:
+                self.aspect_ratio = width / height
+                return
+        
+        # Fallback to bounding box methods
         try:
-            if hasattr(self.annotation, 'rect'):  # RectangleAnnotation
-                rect = self.annotation.rect
-                if rect.height() > 0:
-                    self.aspect_ratio = rect.width() / rect.height()
-            elif hasattr(self.annotation, 'size'):  # PatchAnnotation
-                self.aspect_ratio = 1.0
-            elif hasattr(self.annotation, 'polygon'):  # PolygonAnnotation
-                rect = self.annotation.polygon.boundingRect()
-                if rect.height() > 0:
-                    self.aspect_ratio = rect.width() / rect.height()
-            else:
-                # Fallback for other types or if geometry is not available
-                self.aspect_ratio = 1.0
-        except Exception as e:
-            print(f"Could not determine aspect ratio for {self.annotation.id}: {e}")
-            self.aspect_ratio = 1.0
+            top_left = annotation.get_bounding_box_top_left()
+            bottom_right = annotation.get_bounding_box_bottom_right()
+            
+            if top_left and bottom_right:
+                width = bottom_right.x() - top_left.x()
+                height = bottom_right.y() - top_left.y()
+                
+                if height > 0:
+                    self.aspect_ratio = width / height
+                    return
+        except (AttributeError, TypeError):
+            pass
+        
+        # Last resort: try to get aspect ratio from the cropped image
+        try:
+            pixmap = annotation.get_cropped_image()
+            if pixmap and not pixmap.isNull() and pixmap.height() > 0:
+                self.aspect_ratio = pixmap.width() / pixmap.height()
+                return
+        except (AttributeError, TypeError):
+            pass
+        
+        # Default to square if we can't determine aspect ratio
+        self.aspect_ratio = 1.0
 
     def load_image(self):
         """Loads the image pixmap if it hasn't been loaded yet."""
@@ -356,16 +377,25 @@ class AnnotationDataItem:
         return "<br>".join(tooltip_parts)
 
     def get_effective_confidence(self):
-        """Get the effective confidence value."""
+        """
+        Get the effective confidence value, handling scalar, array, and vector predictions.
+        """
         # First check if prediction probabilities are available from model predictions
         if hasattr(self, 'prediction_probabilities') and self.prediction_probabilities is not None:
-            if len(self.prediction_probabilities) > 0:
-                # Use the maximum probability for confidence sorting
-                return float(np.max(self.prediction_probabilities))
-        
+            probs = self.prediction_probabilities
+            try:
+                # This will succeed for lists and multi-element numpy arrays
+                if len(probs) > 0:
+                    return float(np.max(probs))
+            except TypeError:
+                # This will catch the error if `len()` is called on a scalar or 0-D array.
+                # In this case, the value of `probs` itself is the confidence score.
+                return float(probs)
+
         # Fallback to existing confidence values
         if self.annotation.verified and hasattr(self.annotation, 'user_confidence') and self.annotation.user_confidence:
             return list(self.annotation.user_confidence.values())[0]
         elif hasattr(self.annotation, 'machine_confidence') and self.annotation.machine_confidence:
             return list(self.annotation.machine_confidence.values())[0]
+            
         return 0.0