Add files using upload-large-folder tool

README.md

@@ -1,3 +1,41 @@
----
-license: apache-2.0
----
+---
+pretty_name: ThinkGeoBench Question-4
+license: other
+task_categories:
+- visual-question-answering
+- image-classification
+- tabular-classification
+configs:
+- config_name: default
+  data_files:
+  - split: test
+    path: metadata.jsonl
+---
+
+# ThinkGeoBench Question-4
+
+This repository contains an upload-ready Hugging Face release folder for the `question-4` split of ThinkGeoBench.
+
+## Included files
+
+- `metadata.jsonl`: normalized table for Hugging Face Dataset Viewer
+- `images/`: image assets referenced by `question-4`
+- `assets/AppleSizeEstimate/`: depth `.npy` assets referenced by Apple size estimation samples
+- `question-4.original.json`: original source question file
+- `question_taxonomy_summary.md`: taxonomy and template notes
+
+## Split
+
+- `test`: 539 samples
+
+## Schema notes
+
+- `classification_labels` keeps the original multi-label information as a list.
+- `primary_classification` is the first label, useful for filtering and faceting.
+- `image1_file_name` to `image15_file_name` are relative paths to image assets in this repository.
+- `depth1_path` is a relative path to a `.npy` depth asset when required.
+- `focal_length_px` is present for Apple size estimation samples that require camera intrinsics.
+
+## License note
+
+The placeholder `license: other` should be replaced with the actual redistribution license before publishing publicly.

prepare_summary.json

@@ -0,0 +1,24 @@
+{
+  "source_question_file": "/public/home/yezi/ThinkGeo/my_dataset_test/question-4.json",
+  "output_directory": "/public/home/yezi/ThinkGeo/hf_release_question4",
+  "sample_count": 539,
+  "referenced_file_count": 1137,
+  "referenced_total_bytes": 1027433225,
+  "referenced_total_gib": 0.957,
+  "included_optional_files": [
+    "question-4.original.json",
+    "question_taxonomy_summary.md",
+    "README.md",
+    "metadata.jsonl"
+  ],
+  "excluded_large_or_non_dataset_dirs": [
+    "Evaluator",
+    "tmp",
+    "Statistics",
+    "models",
+    "code",
+    "Papers",
+    "answer_split",
+    "answer_split_v2"
+  ]
+}

question_taxonomy_summary.md

@@ -0,0 +1,221 @@
+# Question Taxonomy Summary
+
+Source file: [question-4.json](question-4.json)
+
+Total questions: `539`
+
+Unique question templates: `76`
+
+Current template IDs present in [question-4.json](question-4.json): `1-76`
+
+Missing template IDs in the current file: `None`
+
+## Current 5-category taxonomy
+
+This version follows the 5-way task taxonomy used for [question-4.json](question-4.json). It is intended for dataset organization, task routing, and template-level traceability. Mixed questions are allowed to belong to more than one category. The template system below is the deduplicated `76-template` version of `question-4`.
+
+1. `识别与检索类 / Recognition & Retrieval`
+   Focus: identify pests, diseases, weeds, or plant taxa, then provide related biological or management information.
+   Typical tasks: pest recognition, crop disease recognition, taxonomic identification, native-range questions, treatment suggestions, causal or ecological relation questions.
+
+2. `目标检测、计数与测量类 / Detection, Counting & Measurement`
+   Focus: detect objects or instances and perform counting, size estimation, diameter measurement, ranking, and coverage estimation at the object level.
+   Typical tasks: tree counting, tree diameter estimation, fruit counting, apple size estimation, wheat-ear counting, lesion counting, and local density comparison.
+
+3. `变化检测类 / Change Detection`
+   Focus: compare two temporal images and identify land-cover or pond changes, changed areas, changed categories, and transition proportions.
+   Typical tasks: cultivated-to-non-cultivated conversion, farm-pond change detection, changed-region localization, and change percentage estimation.
+
+4. `语义分割解译类 / Semantic Segmentation Interpretation`
+   Focus: interpret single-image dense prediction or segmentation-style outputs, including lesion area estimation, weed/crop area proportion, canopy threshold-based region analysis, and mask-based spatial description.
+   Typical tasks: lesion-area percentage, lesion-region localization, weed-vs-crop area interpretation, thresholded canopy coverage, and mask-based regional description.
+
+5. `统计分析型任务 / Statistical Analysis`
+   Focus: perform structured comparison, metric aggregation, band-selection analysis, and curve or chart generation.
+   Typical tasks: multispectral band-combination comparison, grouped metric plots, weed growth curves, and derived summary statistics across options.
+
+## Multi-label rule
+
+The `classification` field in [question-4.json](question-4.json) supports:
+
+- a single integer, such as `1`
+- or an integer array for mixed tasks, such as `[1, 4]`
+
+Use multiple labels when a question has two equally central task components and the answer depends on both.
+
+## Category membership counts
+
+Because multi-label questions are counted once in each category they belong to, the counts below are membership counts rather than a partition of `539`. Template counts here also follow category membership rather than question-4 block placement.
+
+- `Class 1`: `180` memberships, `19` templates
+- `Class 2`: `102` memberships, `22` templates
+- `Class 3`: `103` memberships, `5` templates
+- `Class 4`: `131` memberships, `10` templates
+- `Class 5`: `77` memberships, `29` templates
+
+Mixed-label samples: `54`
+
+Mixed-label templates: `9`
+
+Observed mixed-label combinations: `4`
+
+## question-4.json 分类统计汇总
+
+### 1. 按分类出现次数统计
+
+这里按照“分类出现了多少次”统计。若某条样本是多标签，例如 `[1, 4]`，则会同时计入 `Class 1` 和 `Class 4`。
+
+- `Class 1`: `180`
+- `Class 2`: `102`
+- `Class 3`: `103`
+- `Class 4`: `131`
+- `Class 5`: `77`
+
+总样本数: `539`
+
+多标签样本数: `54`
+
+### 2. 按单标签样本数统计
+
+这里仅统计 `classification` 为单个整数的样本，不把多标签样本重复计入单类。
+
+- `Class 1`: `145`
+- `Class 2`: `90`
+- `Class 3`: `103`
+- `Class 4`: `80`
+- `Class 5`: `67`
+
+单标签样本总数: `485`
+
+### 3. 多标签组合分布
+
+- `[1, 2]`: `3`
+- `[1, 4]`: `32`
+- `[2, 4]`: `9`
+- `[4, 5]`: `10`
+
+### 4. question-4 排列规则与区间
+
+question-4 按 5 个分类块排序，块内再按去重后的上一版 question-4 模板家族升序；混合标签模板按第一个分类标签归块。
+
+- `[1, 4]` questions are placed in the `Class 1` block
+- `[2, 4]` questions are placed in the `Class 2` block
+- `[4, 5]` questions are placed in the `Class 4` block
+
+- `Class 1` block: new templates `1-19`, sample_ids `1-180`, `180` questions, `19` block templates; previous question-4 template families `1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19`
+- `Class 2` block: new templates `20-38`, sample_ids `181-279`, `99` questions, `19` block templates; previous question-4 template families `20, 21, 22/23/24/33/34/35/36, 25, 26, 27, 28/30, 29, 31, 32, 37/38, 39/40/41/42/43, 44, 45, 46, 47, 48, 49, 50`
+- `Class 3` block: new templates `39-43`, sample_ids `280-382`, `103` questions, `5` block templates; previous question-4 template families `51, 52, 53, 54/55, 56`
+- `Class 4` block: new templates `44-48`, sample_ids `383-472`, `90` questions, `5` block templates; previous question-4 template families `57, 58, 59, 60, 61`
+- `Class 5` block: new templates `49-76`, sample_ids `473-539`, `67` questions, `28` block templates; previous question-4 template families `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`
+
+### 5. question-4 中的多标签模板位置
+
+- `[1, 2]`: new templates `8, 9, 10`
+- `[1, 4]`: new templates `11, 12, 13, 14`
+- `[2, 4]`: new templates `38`
+- `[4, 5]`: new templates `46`
+
+## Template Catalog
+
+Each table below follows the final question-4 block order. Multi-label templates appear in the block determined by the first classification label. The provenance columns show both the previous question-4 template family and the original question-3 source template(s).
+
+### Class 1: 识别与检索类 / Recognition & Retrieval
+
+| New template | Previous question-4 template(s) | Question-3 source template(s) | Count | Classification | Reference tool chain | Representative question |
+| --- | --- | --- | --- | --- | --- | --- |
+| `1` | `1` | `1` | `13` | `1` | PestCropIdentification | What agricultural pest is shown in the image? |
+| `2` | `2` | `2` | `12` | `1` | PestCropIdentification | What crop disease is shown in the image? |
+| `3` | `3` | `3` | `15` | `1` | PestCropIdentification, AgriculturalWebSearch | What ecological relationship best describes the interaction between the pests in Image 1 and Image 2: predation, parasitism, competition, mutualism, commensalism, intraspecific cooperation, or none of the above relationships? |
+| `4` | `4` | `4` | `15` | `1` | PestCropIdentification, PestCropIdentification, AgriculturalWebSearch | Is the pest shown in Image 1 a primary cause of the crop disease shown in Image 2? |
+| `5` | `5` | `5` | `20` | `1` | PestCropIdentification, AgriculturalWebSearch | What pesticide is generally most effective against the pest shown in the image, and what control method should be used to manage it effectively? |
+| `6` | `6` | `6` | `20` | `1` | PestCropIdentification, AgriculturalWebSearch | Where is the pest shown in the image currently most widely distributed in the world? |
+| `7` | `7` | `7` | `20` | `1` | PestCropIdentification, AgriculturalWebSearch | Is the crop disease shown in the image infectious or non-infectious, and what is its main cause? |
+| `8` | `8` | `8` | `1` | `[1, 2]` | ExtractMarkedRedBox, ObjectCounting, ImageRegionUnderstanding | How many trees are inside the red box in the picture? Based on the images, are these trees in a healthy growth state? |
+| `9` | `9` | `9` | `1` | `[1, 2]` | ExtractMarkedRedBox, ObjectCounting, ImageRegionUnderstanding | How many trees are inside the red box in the picture? Based on their characteristics in the picture, are they pine trees? |
+| `10` | `10` | `10` | `1` | `[1, 2]` | ExtractMarkedRedBox, ObjectCounting, ImageRegionUnderstanding, AgriculturalWebSearch | How many trees are inside the red box in the picture? In what climatic regions does this type of tree typically grow? |
+| `11` | `11` | `94` | `1` | `[1, 4]` | PestCropIdentification, PlantSegLesionAnalysis, Calculator | What is the condition of the fruit in the picture? Approximately what percentage of the area is infected with disease? |
+| `12` | `12` | `95` | `11` | `[1, 4]` | PestCropIdentification, PlantSegLesionAnalysis, Calculator | What disease is shown in the picture? Approximately what percentage of the affected plant organ area is covered by visible disease lesions? |
+| `13` | `13` | `96` | `10` | `[1, 4]` | PestCropIdentification, PlantSegLesionAnalysis, Calculator, AgriculturalWebSearch | Approximately what percentage of the affected plant organ area is covered by visible disease lesions? What medications or treatments should be used for this disease? |
+| `14` | `14` | `98` | `10` | `[1, 4]` | PestCropIdentification, PlantSegLesionAnalysis, ImageRegionUnderstanding, AgriculturalWebSearch | What disease is shown in the image? Please determine whether it is indeed this disease based on the description of the lesion segmentation result image. |
+| `15` | `15` | `99` | `5` | `1` | WeedSpeciesClassification, AgriculturalWebSearch | Among the three plants shown, exactly one belongs to the family Asteraceae. Which one is it? Respond with its accepted English common name only. |
+| `16` | `16` | `100` | `5` | `1` | WeedSpeciesClassification, AgriculturalWebSearch | Among the three plants shown, exactly one belongs to the family Verbenaceae. Which one is it? Respond with its accepted English common name only. |
+| `17` | `17` | `101` | `5` | `1` | WeedSpeciesClassification, AgriculturalWebSearch | Among the three plants shown, exactly one belongs to the family Fabaceae. Which one is it? Respond with its accepted English common name only. |
+| `18` | `18` | `102` | `5` | `1` | WeedSpeciesClassification, AgriculturalWebSearch | Among the three plants shown, exactly one belongs to the family Rhamnaceae. Which one is it? Respond with its accepted English common name only. |
+| `19` | `19` | `103` | `10` | `1` | WeedSpeciesClassification, AgriculturalWebSearch | What is the native range (natural region of origin) of the weed species shown in the image? Answer in one or two short English sentences. |
+
+### Class 2: 目标检测、计数与测量类 / Detection, Counting & Measurement
+
+| New template | Previous question-4 template(s) | Question-3 source template(s) | Count | Classification | Reference tool chain | Representative question |
+| --- | --- | --- | --- | --- | --- | --- |
+| `20` | `20` | `11` | `2` | `2` | ExtractMarkedRedBox, ObjectCounting, Calculator | What is the diameter of the smallest tree in the red box? GSD 0.08m/pixel |
+| `21` | `21` | `12` | `1` | `2` | ExtractMarkedRedBox, ObjectCounting, Calculator | What is the diameter of the biggest tree in the red box? GSD 0.08m/pixel |
+| `22` | `22,23,24,33,34,35,36` | `13,14,15,24,25,26,27` | `7` | `2` | ExtractMarkedRedBox, ObjectCounting, Calculator | What is the average diameter of the trees in the red box in the picture? GSD 0.08m/pixel |
+| `23` | `25` | `16` | `1` | `2` | ExtractMarkedRedBox, ObjectCounting (with ROI; coordinates do not need to be returned), ObjectCounting (whole image), Calculator | What percentage of the total number of trees in the image is represented by the trees framed in red? |
+| `24` | `26` | `17` | `1` | `2` | ExtractMarkedRedBox, ObjectCounting, Calculator | What is the difference in diameter between the largest and smallest trees within the red box? GSD 0.08m/pixel |
+| `25` | `27` | `18` | `1` | `2` | ExtractMarkedRedBox, ObjectCounting, Calculator | What is the standard deviation of the diameter of the trees in the red box in the picture? |
+| `26` | `28,30` | `19,21` | `2` | `2` | ExtractMarkedRedBox, ObjectCounting | How many trees are inside the red box in the picture? |
+| `27` | `29` | `20` | `15` | `2` | ExtractMarkedRedBox, ObjectCounting (ROI is the red-box region), ObjectCounting (pass in the inferred ROI for this column), Calculator | How many trees are in the red box in the picture? What percentage of the total number of trees in this column are represented? |
+| `28` | `31` | `22` | `1` | `2` | ExtractMarkedRedBox, ObjectCounting, Calculator | What are the diameters of the trees within the red box? GSD 0.08m/pixel |
+| `29` | `32` | `23` | `1` | `2` | ExtractMarkedRedBox, Calculator (compute the red-box area), ObjectCounting (obtain tree coordinates), Calculator (compute the total tree area), Calculator (compute the percentage) | What is the area within the red box in the image? Approximately what percentage is covered by trees? |
+| `30` | `37,38` | `28,29` | `14` | `2` | ExtractMarkedRedBox, AppleSizeEstimate (pass the red-box region as ROI), AppleSizeEstimate (whole image) | Where does the diameter of the apple within the red box rank among the total size of all apples in the image? How big is it? |
+| `31` | `39,40,41,42,43` | `30,31,32,33,34` | `15` | `2` | ExtractMarkedRedBox, AppleSizeEstimate (pass the red-box region as ROI), AppleSizeEstimate (whole image) | Is the apple in the red box the largest apple in the whole picture? How big is it? |
+| `32` | `44` | `35` | `11` | `2` | AppleSizeEstimate (whole image), Calculator | What is the average size of the apples in the picture? |
+| `33` | `45` | `36` | `8` | `2` | ObjectCounting (whole image; correctly crop the image into a 3x3 grid before calling) | Divide the image into 9 equal squares, with the top left square being the first and the bottom right square being the ninth. Which square has the most oranges? How many exactly? |
+| `34` | `46` | `37` | `3` | `2` | ObjectCounting (Image 1), ObjectCounting (Image 2) | Which picture contains more blueberries? By how much? |
+| `35` | `47` | `39` | `3` | `2` | ObjectCounting | How many wheat ears are there in the picture? |
+| `36` | `48` | `40` | `3` | `2` | ObjectCounting (whole image; correctly crop the image into a 3x3 grid before calling) | Divide the image into 9 equal squares, with the top left square being the first and the bottom right square being the ninth. Which square has the most wheat ears? How many exactly? |
+| `37` | `49` | `41` | `1` | `2` | ObjectCounting (Image 1), ObjectCounting (Image 2) | Which picture contains more wheat ears? By how much? |
+| `38` | `50` | `97` | `9` | `[2, 4]` | PlantSegLesionAnalysis, ImageRegionUnderstanding | How many lesions are there in the crop shown in the picture? Please describe where the most severe lesion areas are concentrated based on the segmented lesion images. |
+
+### Class 3: 变化检测类 / Change Detection
+
+| New template | Previous question-4 template(s) | Question-3 source template(s) | Count | Classification | Reference tool chain | Representative question |
+| --- | --- | --- | --- | --- | --- | --- |
+| `39` | `51` | `42` | `59` | `3` | ChangeDetection, Calculator | From Figure 1 to Figure 2, what percentage of the total area of ​​the map is converted from cultivated land to non-cultivated land? |
+| `40` | `52` | `43` | `10` | `3` | ExtractMarkedRedBox, ChangeDetection, Calculator | Has any farmland in the area marked by the red box in the map been converted to non-farmland? If so, what percentage of the total area of ​​the map does it represent? |
+| `41` | `53` | `44` | `19` | `3` | ChangeDetection, Calculator | How many types of farm pond changes occurred from Figure 1 to Figure 2? What are their respective proportions of the total area in the images? |
+| `42` | `54,55` | `45,46` | `7` | `3` | ChangeDetection, ImageRegionUnderstanding | Based on the final masked image, describe the location of the changed pond types in the image (if the image is divided into nine grids, where are they located?). |
+| `43` | `56` | `47` | `8` | `3` | ChangeDetection, Calculator | Are there any changes to farm ponds in the two images? What percentage of the entire image represents the changes? |
+
+### Class 4: 语义分割解译类 / Semantic Segmentation Interpretation
+
+| New template | Previous question-4 template(s) | Question-3 source template(s) | Count | Classification | Reference tool chain | Representative question |
+| --- | --- | --- | --- | --- | --- | --- |
+| `44` | `57` | `48` | `10` | `4` | CanopyHeightEstimate (pass in 2 m), Calculator | If vegetation in the diagram that is taller than 2m is considered to be trees, what is the tree coverage rate in the diagram? |
+| `45` | `58` | `49` | `20` | `4` | CanopyHeightEstimate (4 m), CanopyHeightEstimate (2 m), Calculator (compute pixels below 2 m), Calculator (compute the difference) | Is the proportion of vegetation areas with a height greater than 4m larger or smaller than that of vegetation areas with a height less than 2m in the overall map? Specifically, by how much? |
+| `46` | `59` | `50` | `20` | `[4, 5]` | CanopyHeightEstimate (0 m to inspect all image pixels and the maximum height), Calculator (compute the threshold by taking 10% down from the maximum vegetation height), CanopyHeightEstimate (using the threshold from the previous step), Calculator (compute the percentage) | Among all vegetation on the map, what percentage of the total map area is occupied by regions with vegetation height in the top 10%? |
+| `47` | `60` | `52` | `35` | `4` | WeedSegmentationAnalysis, Calculator (compute the weed/crop proportions over the whole image), Calculator (compute the weed/crop proportions among all non-background pixels) | In this image, do weeds or crops occupy a larger proportion of the whole image, and by how much? Among all non-background pixels, what percentage is weeds and what percentage is crops? |
+| `48` | `61` | `104` | `5` | `4` | CanopyHeightEstimate (2 m), CanopyHeightEstimate (4 m), CanopyHeightEstimate (6 m), Calculator (compute the 2-4 m percentage over the whole image), Calculator (compute the 4-6 m percentage over the whole image) | Among all image pixels, what percentage of the total image area is occupied by vegetation with heights between 2 m and 4 m, and what percentage is occupied by vegetation with heights between 4 m and 6 m? |
+
+### Class 5: 统计分析型任务 / Statistical Analysis
+
+| New template | Previous question-4 template(s) | Question-3 source template(s) | Count | Classification | Reference tool chain | Representative question |
+| --- | --- | --- | --- | --- | --- | --- |
+| `49` | `62` | `51` | `40` | `5` | WeedPhenotypingAnalysis, Plot (follow the requirements in the question) | Plot the weed height curves from the images above: put week on the horizontal axis and height (cm) on the vertical axis, and draw one colored polyline per species. It also needs to include a legend, horizontal and vertical axis scales, and names. |
+| `50` | `63` | `54` | `1` | `5` | WeedSegmentationAnalysis, WeedSegmentationAnalysis, WeedSegmentationAnalysis | Among the three multispectral band-combination options below, which one achieves the **highest mean IoU (averaged across classes)** on this tile?<br><br>- **A**: [RGB]<br>- **B**: [NIR,RGB]<br>- **C**: [NDVI] |
+| `51` | `64` | `55` | `1` | `5` | WeedSegmentationAnalysis, WeedSegmentationAnalysis, WeedSegmentationAnalysis | Among the three multispectral band-combination options below, which one achieves the **highest weed IoU** on this tile?<br><br>- **A**: [RGB]<br>- **B**: [G,NIR,R,RGB]<br>- **C**: [NDVI] |
+| `52` | `65` | `56` | `1` | `5` | WeedSegmentationAnalysis, WeedSegmentationAnalysis, WeedSegmentationAnalysis | Among the three multispectral band-combination options below, which one achieves the **highest crop IoU** on this tile?<br><br>- **A**: [NIR]<br>- **B**: [B,G,NIR,R,RE,RGB]<br>- **C**: [B,CIR,G,NIR,R,RE,RGB] |
+| `53` | `66` | `57` | `1` | `5` | WeedSegmentationAnalysis, WeedSegmentationAnalysis, WeedSegmentationAnalysis | Among the three multispectral band-combination options below, which one achieves the **highest overall accuracy** on this tile?<br><br>- **A**: [RGB]<br>- **B**: [NDVI]<br>- **C**: [NIR] |
+| `54` | `67` | `58` | `1` | `5` | WeedSegmentationAnalysis, WeedSegmentationAnalysis, WeedSegmentationAnalysis | Among the three multispectral band-combination options below, which one achieves the **highest weed F1 score** on this tile?<br><br>- **A**: [NIR,RGB]<br>- **B**: [RGB]<br>- **C**: [NDVI] |
+| `55` | `68` | `59` | `1` | `5` | WeedSegmentationAnalysis, WeedSegmentationAnalysis, WeedSegmentationAnalysis | Among the three multispectral band-combination options below, which one achieves the **highest mean F1 score** on this tile?<br><br>- **A**: [B,CIR,G,NDVI,NIR,R,RE,RGB]<br>- **B**: [RGB]<br>- **C**: [NDVI] |
+| `56` | `69` | `60` | `1` | `5` | WeedSegmentationAnalysis, WeedSegmentationAnalysis, WeedSegmentationAnalysis | Among the three multispectral band-combination options below, which one achieves the **highest mean PR-AUC** on this tile?<br><br>- **A**: [B,CIR,G,NDVI,NIR,R,RE,RGB]<br>- **B**: [RGB]<br>- **C**: [NIR,RGB] |
+| `57` | `70` | `61` | `1` | `5` | WeedSegmentationAnalysis, WeedSegmentationAnalysis, WeedSegmentationAnalysis | Among the three multispectral band-combination options below, which one achieves the **highest crop F1 score** on this tile?<br><br>- **A**: [RGB]<br>- **B**: [NDVI]<br>- **C**: [B,G,NDVI,NIR,R,RE,RGB] |
+| `58` | `71` | `62` | `1` | `5` | WeedSegmentationAnalysis, WeedSegmentationAnalysis, WeedSegmentationAnalysis | Among the three multispectral band-combination options below, which one achieves the **highest frequency-weighted IoU** on this tile?<br><br>- **A**: [RGB]<br>- **B**: [NDVI]<br>- **C**: [NIR,RGB] |
+| `59` | `72` | `63` | `1` | `5` | WeedSegmentationAnalysis, WeedSegmentationAnalysis, WeedSegmentationAnalysis | Among the three multispectral band-combination options below, which one achieves the **highest mean accuracy** on this tile?<br><br>- **A**: [G,NIR,R,RGB]<br>- **B**: [RGB]<br>- **C**: [NDVI] |
+| `60` | `73` | `64` | `1` | `5` | WeedSegmentationAnalysis, WeedSegmentationAnalysis, WeedSegmentationAnalysis | Among the three multispectral band-combination options below, which one achieves the **highest crop precision** on this tile?<br><br>- **A**: [B,CIR,G,NDVI,NIR,R,RE]<br>- **B**: [RGB]<br>- **C**: [NIR,RGB] |
+| `61` | `74` | `65` | `1` | `5` | WeedSegmentationAnalysis, WeedSegmentationAnalysis, WeedSegmentationAnalysis | Among the three multispectral band-combination options below, which one achieves the **highest weed precision** on this tile?<br><br>- **A**: [RGB]<br>- **B**: [NIR]<br>- **C**: [NDVI] |
+| `62` | `75` | `66` | `1` | `5` | WeedSegmentationAnalysis, WeedSegmentationAnalysis, WeedSegmentationAnalysis | Among the three multispectral band-combination options below, which one achieves the **highest mean average precision (MAP)** on this tile?<br><br>- **A**: [RGB]<br>- **B**: [NDVI]<br>- **C**: [B,G,NIR,R,RE,RGB] |
+| `63` | `76` | `67` | `1` | `5` | WeedSegmentationAnalysis, WeedSegmentationAnalysis, WeedSegmentationAnalysis | Among the three multispectral band-combination options below, which one achieves the **highest crop accuracy** on this tile?<br><br>- **A**: [NIR,RGB]<br>- **B**: [RGB]<br>- **C**: [G,NIR,R,RGB] |
+| `64` | `77` | `68` | `1` | `5` | WeedSegmentationAnalysis, WeedSegmentationAnalysis, WeedSegmentationAnalysis | Among the three multispectral band-combination options below, which one achieves the **highest weed accuracy** on this tile?<br><br>- **A**: [RGB]<br>- **B**: [NDVI]<br>- **C**: [B,G,NDVI,NIR,R,RE,RGB] |
+| `65` | `78` | `69` | `1` | `5` | WeedSegmentationAnalysis, Plot | The RGB preview of one field tile is shown. Three **candidate band-combination presets** are:<br><br>- **A**: [RGB]<br>- **B**: [NIR,RGB]<br>- **C**: [B,CIR,G,NIR,R,RE,RGB]<br><br>For each preset, quantify **mean accuracy**, **mean precision**, **mean F1**, and **mean IoU** on this tile. Present all **four** metrics **in one figure**: a **grouped bar chart** where the **four metrics** share one axis and **values** lie on the other; use **distinct colors** for presets A, B, and C, include a **legend**, and give both axes **clear titles and tick marks**. |
+| `66` | `79` | `70` | `1` | `5` | WeedSegmentationAnalysis, Plot | The RGB preview of one field tile is shown. Three **candidate band-combination presets** are:<br><br>- **A**: [NDVI]<br>- **B**: [RGB]<br>- **C**: [B,G,NIR,R,RE,RGB]<br><br>For each preset, quantify **mean accuracy**, **mean precision**, **mean F1**, and **mean IoU** on this tile. Present all **four** metrics **in one figure**: a **grouped bar chart** where the **four metrics** share one axis and **values** lie on the other; use **distinct colors** for presets A, B, and C, include a **legend**, and give both axes **clear titles and tick marks**. |
+| `67` | `80` | `71` | `1` | `5` | WeedSegmentationAnalysis, Plot | The RGB preview of one field tile is shown. Three **candidate band-combination presets** are:<br><br>- **A**: [B,CIR,G,NDVI,NIR,R,RE,RGB]<br>- **B**: [RGB]<br>- **C**: [G,NIR,R,RGB]<br><br>For each preset, quantify **mean accuracy**, **mean precision**, **mean F1**, and **mean IoU** on this tile. Present all **four** metrics **in one figure**: a **grouped bar chart** where the **four metrics** share one axis and **values** lie on the other; use **distinct colors** for presets A, B, and C, include a **legend**, and give both axes **clear titles and tick marks**. |
+| `68` | `81` | `72` | `1` | `5` | WeedSegmentationAnalysis, Plot | The RGB preview of one field tile is shown. Three **candidate band-combination presets** are:<br><br>- **A**: [RGB]<br>- **B**: [NDVI]<br>- **C**: [NIR]<br><br>For each preset, quantify **mean accuracy**, **mean precision**, **mean F1**, and **mean IoU** on this tile. Present all **four** metrics **in one figure**: a **grouped bar chart** where the **four metrics** share one axis and **values** lie on the other; use **distinct colors** for presets A, B, and C, include a **legend**, and give both axes **clear titles and tick marks**. |
+| `69` | `82` | `73` | `1` | `5` | WeedSegmentationAnalysis, Plot | The RGB preview of one field tile is shown. Three **candidate band-combination presets** are:<br><br>- **A**: [NIR,RGB]<br>- **B**: [RGB]<br>- **C**: [B,G,NDVI,NIR,R,RE,RGB]<br><br>For each preset, quantify **crop** accuracy, **crop** precision, **crop** F1, and **crop** IoU on this tile. Present all **four** metrics **in one figure**: a **grouped bar chart** where the **four metrics** share one axis and **values** lie on the other; use **distinct colors** for presets A, B, and C, include a **legend**, and give both axes **clear titles and tick marks**. |
+| `70` | `83` | `74` | `1` | `5` | WeedSegmentationAnalysis, Plot | The RGB preview of one field tile is shown. Three **candidate band-combination presets** are:<br><br>- **A**: [B,CIR,G,NIR,R,RE,RGB]<br>- **B**: [RGB]<br>- **C**: [NDVI]<br><br>For each preset, quantify **crop** accuracy, **crop** precision, **crop** F1, and **crop** IoU on this tile. Present all **four** metrics **in one figure**: a **grouped bar chart** where the **four metrics** share one axis and **values** lie on the other; use **distinct colors** for presets A, B, and C, include a **legend**, and give both axes **clear titles and tick marks**. |
+| `71` | `84` | `75` | `1` | `5` | WeedSegmentationAnalysis, Plot | The RGB preview of one field tile is shown. Three **candidate band-combination presets** are:<br><br>- **A**: [RGB]<br>- **B**: [B,CIR,G,NDVI,NIR,R,RE]<br>- **C**: [G,NIR,R,RGB]<br><br>For each preset, quantify **crop** accuracy, **crop** precision, **crop** F1, and **crop** IoU on this tile. Present all **four** metrics **in one figure**: a **grouped bar chart** where the **four metrics** share one axis and **values** lie on the other; use **distinct colors** for presets A, B, and C, include a **legend**, and give both axes **clear titles and tick marks**. |
+| `72` | `85` | `76` | `1` | `5` | WeedSegmentationAnalysis, Plot | The RGB preview of one field tile is shown. Three **candidate band-combination presets** are:<br><br>- **A**: [B,CIR,G,NDVI,NIR,R,RE,RGB]<br>- **B**: [NIR]<br>- **C**: [RGB]<br><br>For each preset, quantify **crop** accuracy, **crop** precision, **crop** F1, and **crop** IoU on this tile. Present all **four** metrics **in one figure**: a **grouped bar chart** where the **four metrics** share one axis and **values** lie on the other; use **distinct colors** for presets A, B, and C, include a **legend**, and give both axes **clear titles and tick marks**. |
+| `73` | `86` | `77` | `1` | `5` | WeedSegmentationAnalysis, Plot | The RGB preview of one field tile is shown. Three **candidate band-combination presets** are:<br><br>- **A**: [RGB]<br>- **B**: [NDVI]<br>- **C**: [G,NIR,R,RGB]<br><br>For each preset, quantify **weed** accuracy, **weed** precision, **weed** F1, and **weed** IoU on this tile. Present all **four** metrics **in one figure**: a **grouped bar chart** where the **four metrics** share one axis and **values** lie on the other; use **distinct colors** for presets A, B, and C, include a **legend**, and give both axes **clear titles and tick marks**. |
+| `74` | `87` | `78` | `1` | `5` | WeedSegmentationAnalysis, Plot | The RGB preview of one field tile is shown. Three **candidate band-combination presets** are:<br><br>- **A**: [NIR]<br>- **B**: [RGB]<br>- **C**: [NIR,RGB]<br><br>For each preset, quantify **weed** accuracy, **weed** precision, **weed** F1, and **weed** IoU on this tile. Present all **four** metrics **in one figure**: a **grouped bar chart** where the **four metrics** share one axis and **values** lie on the other; use **distinct colors** for presets A, B, and C, include a **legend**, and give both axes **clear titles and tick marks**. |
+| `75` | `88` | `79` | `1` | `5` | WeedSegmentationAnalysis, Plot | The RGB preview of one field tile is shown. Three **candidate band-combination presets** are:<br><br>- **A**: [NDVI]<br>- **B**: [RGB]<br>- **C**: [B,CIR,G,NIR,R,RE,RGB]<br><br>For each preset, quantify **weed** accuracy, **weed** precision, **weed** F1, and **weed** IoU on this tile. Present all **four** metrics **in one figure**: a **grouped bar chart** where the **four metrics** share one axis and **values** lie on the other; use **distinct colors** for presets A, B, and C, include a **legend**, and give both axes **clear titles and tick marks**. |
+| `76` | `89` | `80` | `1` | `5` | WeedSegmentationAnalysis, Plot | The RGB preview of one field tile is shown. Three **candidate band-combination presets** are:<br><br>- **A**: [RGB]<br>- **B**: [B,G,NDVI,NIR,R,RE,RGB]<br>- **C**: [NIR,RGB]<br><br>For each preset, quantify **weed** accuracy, **weed** precision, **weed** F1, and **weed** IoU on this tile. Present all **four** metrics **in one figure**: a **grouped bar chart** where the **four metrics** share one axis and **values** lie on the other; use **distinct colors** for presets A, B, and C, include a **legend**, and give both axes **clear titles and tick marks**. |