## Image Comparison: Text Reconstruction Methods ### Overview The image presents a visual comparison of different methods for reconstructing text in street-view images. It shows the results of five different approaches: "Hierarchical-GS", "Hierarchical-GS (T2)", "Our-3D-GS", "Our-Scaffold-GS", and "GT" (Ground Truth). Each method is applied to two different scenes, displayed in two rows. The images are annotated with colored bounding boxes highlighting the reconstructed text regions. ### Components/Axes The image is structured as a grid with two rows and five columns. Each column represents a different text reconstruction method. The rows represent different scenes. The methods are labeled at the top of each column. The bounding boxes are colored red, green, or yellow, depending on the method. ### Detailed Analysis or ### Content Details **Column 1: Hierarchical-GS** * **Top Row:** A street scene with a car parked on the side. A red bounding box highlights the reconstructed text on the car's license plate area. A small red arrow points to a spot on the road. A small red bounding box highlights the text on a sign on the building. * Text in red box on car: "DANS" * **Bottom Row:** A building facade with a scooter parked in front. Red bounding boxes highlight the reconstructed text on the building's architectural details and a sign. **Column 2: Hierarchical-GS (T2)** * **Top Row:** Similar street scene as in Column 1. A red bounding box highlights the reconstructed text on the car's license plate area. A small red bounding box highlights the text on a sign on the building. * **Bottom Row:** Similar building facade as in Column 1. Red bounding boxes highlight the reconstructed text on the building's architectural details and a sign. **Column 3: Our-3D-GS** * **Top Row:** Similar street scene as in Column 1. A red bounding box highlights the reconstructed text on the car's license plate area. A small red bounding box highlights the text on a sign on the building. * Text in red box on car: "ENAGE DANS" * **Bottom Row:** Similar building facade as in Column 1. Red bounding boxes highlight the reconstructed text on the building's architectural details and a sign. **Column 4: Our-Scaffold-GS** * **Top Row:** Similar street scene as in Column 1. A green bounding box highlights the reconstructed text on the car's license plate area. * Text in green box on car: "BRAYA FINAGE DANS" * **Bottom Row:** Similar building facade as in Column 1. Green bounding boxes highlight the reconstructed text on the building's architectural details and a sign. **Column 5: GT (Ground Truth)** * **Top Row:** Similar street scene as in Column 1. A yellow bounding box highlights the reconstructed text on the car's license plate area. A yellow bounding box highlights the text on a sign on the building. * Text in yellow box on car: "BRAYA EINAGE DANS" * **Bottom Row:** Similar building facade as in Column 1. Yellow bounding boxes highlight the reconstructed text on the building's architectural details and a sign. ### Key Observations * The "GT" column represents the ground truth, showing the ideal text reconstruction. * The different methods show varying degrees of success in reconstructing the text. * "Our-Scaffold-GS" appears to produce results closer to the ground truth compared to "Hierarchical-GS" and "Our-3D-GS". * The red arrow in the first image of "Hierarchical-GS" does not appear to be related to text reconstruction. ### Interpretation The image provides a visual comparison of different text reconstruction methods in street-view images. The goal is to assess the accuracy and effectiveness of each method in recovering text from real-world scenes. The "GT" column serves as a benchmark for evaluating the performance of the other methods. The results suggest that "Our-Scaffold-GS" performs better than "Hierarchical-GS" and "Our-3D-GS" in the specific scenes depicted. The differences in performance likely stem from the underlying algorithms and assumptions of each method. The image highlights the challenges of text reconstruction in complex environments and the importance of developing robust and accurate algorithms.

\n ## Image Series: Object Detection Results ### Overview The image presents a series of five views of the same street scene, demonstrating the results of different object detection algorithms. Each view shows the same scene with bounding boxes highlighting detected objects. The algorithms being compared are "Hierarchical-GS", "Hierarchical-GS (τ₂)", "Our-3D-GS", "Our-Scaffold-GS", and "GT" (Ground Truth). Two distinct scenes are shown, one with a car and the other with a motorcycle. ### Components/Axes The image is organized into a 2x5 grid. Each column represents a different algorithm's output. The rows show two different scenes. The bounding boxes are color-coded: * **Red:** Used by "Hierarchical-GS" and "Hierarchical-GS (τ₂)" * **Green:** Used by "Our-3D-GS" and "Our-Scaffold-GS" * **Yellow:** Used by "GT" The top row focuses on a black car, and the bottom row focuses on a motorcycle. Text labels are present on signs in the scene, some of which are partially visible. ### Detailed Analysis or Content Details **Top Row (Car Scene):** * **Hierarchical-GS:** A red bounding box surrounds the black car. The box appears to accurately encompass the vehicle. * **Hierarchical-GS (τ₂):** A red bounding box surrounds the black car, similar to the previous algorithm. * **Our-3D-GS:** A red bounding box surrounds the black car. * **Our-Scaffold-GS:** A green bounding box surrounds the black car. * **GT:** A yellow bounding box surrounds the black car. A sign is visible in the background, with the text "BRAYA IMAGE DANS" (French for "BRAVE IMAGE IN"). **Bottom Row (Motorcycle Scene):** * **Hierarchical-GS:** A red bounding box surrounds the motorcycle. * **Hierarchical-GS (τ₂):** A red bounding box surrounds the motorcycle. * **Our-3D-GS:** A green bounding box surrounds the motorcycle. * **Our-Scaffold-GS:** A green bounding box surrounds the motorcycle. * **GT:** A yellow bounding box surrounds the motorcycle. A small object (possibly a trash can or a small box) is also highlighted with a yellow bounding box in the "GT" image. ### Key Observations * All algorithms successfully detect the primary objects (car and motorcycle) in both scenes. * The "GT" image provides a more complete detection, including the smaller object in the motorcycle scene. * The color-coding allows for a direct visual comparison of the algorithms' performance. * The text on the sign is consistent across all images, indicating the scene remains unchanged. ### Interpretation This image series is a comparative analysis of object detection algorithms. The "GT" (Ground Truth) serves as the benchmark for accurate detection. The other algorithms are evaluated based on their ability to match the "GT" bounding boxes. The consistent detection of the car and motorcycle across all algorithms suggests a reasonable level of performance. The inclusion of the smaller object in the "GT" image highlights the potential for more detailed and comprehensive detection with a more refined ground truth. The use of different colors for each algorithm facilitates a quick visual assessment of their strengths and weaknesses. The French text on the sign is irrelevant to the object detection task but confirms the scene's location or origin. The algorithms appear to be performing similarly, with the main difference being the inclusion of smaller objects in the ground truth.

## Visual Comparison Diagram: 3D Gaussian Splatting Reconstruction Quality ### Overview This image is a qualitative comparison of different 3D Gaussian Splatting (GS) methods for novel view synthesis. It presents a side-by-side visual evaluation across two different scenes (top row: street scene, bottom row: building facade). The comparison aims to demonstrate the visual fidelity and detail preservation of the proposed methods ("Our-3D-GS" and "Our-Scaffold-GS") against baseline methods ("Hierarchical-GS" and "Hierarchical-GS (T2)") and the Ground Truth ("GT"). ### Components/Axes * **Structure:** A 2x5 grid of sub-images. * **Column Headers (Method Labels):** Centered above each column. 1. `Hierarchical-GS` 2. `Hierarchical-GS (T2)` 3. `Our-3D-GS` 4. `Our-Scaffold-GS` 5. `GT` (Ground Truth) * **Visual Annotations:** Colored bounding boxes highlight specific regions of interest for comparison. * **Red Boxes:** Used for `Hierarchical-GS`, `Hierarchical-GS (T2)`, and `Our-3D-GS`. * **Green Box:** Used for `Our-Scaffold-GS`. * **Yellow Box:** Used for `GT`. * **Scenes:** * **Top Row:** A street view with parked cars, buildings, and a prominent sign on a car's rear window. * **Bottom Row:** A close-up view of a building facade with windows, a scooter, and architectural details. ### Detailed Analysis **Top Row - Street Scene:** * **Focus Area:** A sign on the rear window of a dark blue car. * **Text Transcription (Visible in GT):** The sign contains French text. The clearest words are "BRAYA", "INAGE", and "DANS". The full text is partially obscured but appears to be an advertisement or notice. * **Method Comparison (Left to Right):** * `Hierarchical-GS`: The text within the red box is heavily blurred and illegible. * `Hierarchical-GS (T2)`: The text is extremely blurred, appearing as a smudge with no discernible characters. * `Our-3D-GS`: The text is clearer than the previous two but still blurry. Some letter shapes are vaguely visible. * `Our-Scaffold-GS`: The text within the green box is significantly sharper. The words "BRAYA", "INAGE", and "DANS" are readable, though not perfectly crisp. * `GT`: The text within the yellow box is sharp and fully legible, serving as the reference. **Bottom Row - Building Facade:** * **Focus Areas:** Two regions are highlighted: a window on the left and a section of the facade/awning on the right. * **Method Comparison (Left to Right):** * `Hierarchical-GS`: Both red-boxed regions are very blurry. The window pane details and facade texture are lost. * `Hierarchical-GS (T2)`: Similar severe blurriness as the first column. * `Our-3D-GS`: Moderate improvement. Some structural lines are visible, but fine details and textures remain smeared. * `Our-Scaffold-GS`: Notable improvement in the green-boxed regions. The window frame and the vertical lines on the facade are much sharper and more defined, approaching the GT. * `GT`: The yellow-boxed regions show crisp edges, clear window panes, and distinct architectural details. ### Key Observations 1. **Progressive Improvement:** There is a clear visual trend of improving reconstruction quality from left to right across the columns, culminating in the `GT`. 2. **Text as a Key Differentiator:** The ability to reconstruct legible text (top row) is a strong differentiator. `Our-Scaffold-GS` performs markedly better than the Hierarchical baselines and `Our-3D-GS` in this regard. 3. **Detail Preservation:** The bottom row demonstrates that `Our-Scaffold-GS` preserves high-frequency details (edges, lines, textures) much better than the other non-GT methods, which produce smoothed-out or blurred results. 4. **Failure Case of `Hierarchical-GS (T2)`:** The `(T2)` variant appears to perform worse than the standard `Hierarchical-GS` in these examples, producing the most blurred results. ### Interpretation This diagram serves as visual evidence for a research paper, arguing for the superiority of the authors' proposed methods, particularly `Our-Scaffold-GS`. The comparison is designed to show that their approach better handles challenging aspects of scene reconstruction: * **Semantic Detail:** Legible text is a high-level semantic feature. The success of `Our-Scaffold-GS` here suggests it better integrates or preserves features critical for recognition. * **Geometric Fidelity:** The sharp edges and lines in the building facade (bottom row) indicate better geometric accuracy and less "floaters" or artifacts common in neural rendering. * **Methodological Progress:** The progression from `Hierarchical-GS` to `Our-3D-GS` to `Our-Scaffold-GS` implies an iterative improvement in the underlying algorithm, with the scaffold-based approach yielding the most visually convincing results closest to ground truth. The use of colored boxes strategically draws the viewer's eye to the most telling differences, making the argument visually intuitive.

## Comparison of Generative Stereo Methods ### Overview The image presents a side-by-side comparison of four generative stereo (GS) methods applied to urban street scenes, alongside ground truth (GT) images. Each method's output is annotated with colored bounding boxes (red, green, yellow) to highlight regions of interest or discrepancies. The comparison is structured in two rows, with the top row focusing on a street scene with vehicles and the bottom row on a building facade with a motorcycle. ### Components/Axes - **Labels**: - Top row: "Hierarchical-GS," "Hierarchical-GS (τ2)," "Our-3D-GS," "Our-Scaffold-GS," "GT." - Bottom row: Same labels as the top row. - **Annotations**: - **Red boxes**: Highlight regions where the method's output appears superior to GT (e.g., sharper textures, clearer details). - **Green/yellow boxes**: Indicate regions where the method's output diverges from GT (e.g., blurring, artifacts, missing details). - **Text in Boxes**: - Top row: "BRAVA INJAGE DANS" (inside red/yellow boxes). - Bottom row: No explicit text in boxes, but annotations focus on structural details (e.g., building windows, motorcycle). ### Detailed Analysis - **Top Row (Street Scene)**: - **Hierarchical-GS**: Red box on the rear of a dark car; GT has a yellow box in the same area. - **Hierarchical-GS (τ2)**: Red box on the car's rear; GT has a yellow box. - **Our-3D-GS**: Red box on the car's rear; GT has a yellow box. - **Our-Scaffold-GS**: Green box on the car's rear; GT has a yellow box. - **GT**: Yellow box on the car's rear, labeled "BRAVA INJAGE DANS." - **Bottom Row (Building Facade)**: - **Hierarchical-GS**: Red box on the building's wall; GT has a yellow box. - **Hierarchical-GS (τ2)**: Red box on the wall; GT has a yellow box. - **Our-3D-GS**: Red box on the wall; GT has a yellow box. - **Our-Scaffold-GS**: Green box on the wall; GT has a yellow box. - **GT**: Yellow box on the wall, highlighting structural details. ### Key Observations 1. **Red Boxes**: Consistently appear in the first three methods (Hierarchical-GS, Hierarchical-GS (τ2), Our-3D-GS), suggesting these methods preserve certain details (e.g., car textures, building edges) better than GT in specific regions. 2. **Green/Yellow Boxes**: Dominant in "Our-Scaffold-GS," indicating significant discrepancies in texture or structure compared to GT. 3. **GT Annotations**: Yellow boxes in GT images highlight ground truth details (e.g., "BRAVA INJAGE DANS" text), serving as a reference for evaluating method performance. ### Interpretation The comparison demonstrates that: - **Hierarchical-GS and τ2** methods show moderate alignment with GT, with red boxes indicating localized improvements (e.g., sharper car details). - **Our-3D-GS** performs similarly to Hierarchical methods but with slightly fewer red boxes, suggesting comparable but less consistent performance. - **Our-Scaffold-GS** exhibits the most divergence from GT, as evidenced by green/yellow boxes, potentially due to over-smoothing or artifact introduction. - The GT annotations ("BRAVA INJAGE DANS") confirm that the methods are evaluated against real-world text and structural details, emphasizing the importance of fidelity in urban scenes. This analysis underscores the trade-offs between different GS approaches, with some methods excelling in specific regions while others introduce artifacts. The red/green/yellow annotations provide a visual guide to method strengths and weaknesses, critical for refining generative stereo algorithms.