Duplicate from Irikos/RipVIS

Co-authored-by: Andrei Dumitriu <Irikos@users.noreply.huggingface.co>

.gitattributes

@@ -0,0 +1,62 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.lz4 filter=lfs diff=lfs merge=lfs -text
+*.mds filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+# Audio files - uncompressed
+*.pcm filter=lfs diff=lfs merge=lfs -text
+*.sam filter=lfs diff=lfs merge=lfs -text
+*.raw filter=lfs diff=lfs merge=lfs -text
+# Audio files - compressed
+*.aac filter=lfs diff=lfs merge=lfs -text
+*.flac filter=lfs diff=lfs merge=lfs -text
+*.mp3 filter=lfs diff=lfs merge=lfs -text
+*.ogg filter=lfs diff=lfs merge=lfs -text
+*.wav filter=lfs diff=lfs merge=lfs -text
+# Image files - uncompressed
+*.bmp filter=lfs diff=lfs merge=lfs -text
+*.gif filter=lfs diff=lfs merge=lfs -text
+*.png filter=lfs diff=lfs merge=lfs -text
+*.tiff filter=lfs diff=lfs merge=lfs -text
+# Image files - compressed
+*.jpg filter=lfs diff=lfs merge=lfs -text
+*.jpeg filter=lfs diff=lfs merge=lfs -text
+*.webp filter=lfs diff=lfs merge=lfs -text
+# Video files - compressed
+*.mp4 filter=lfs diff=lfs merge=lfs -text
+*.webm filter=lfs diff=lfs merge=lfs -text
+train/coco_annotations/train_with_additional_data.json filter=lfs diff=lfs merge=lfs -text
+train/coco_annotations/train.json filter=lfs diff=lfs merge=lfs -text
+val/coco_annotations/val.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,232 @@
+---
+license: cc-by-nc-4.0
+language:
+- en
+pretty_name: RipVIS
+tags:
+- instance segmentation
+- segmentation
+- computer vision
+- rip
+- rip_current
+size_categories:
+- 100K<n<1M
+---
+# RipVIS v1.8.4
+This Readme describes the RipVIS dataset, its contents, structure, known limitations, how to use it and what to expect in future updates. For more details, future challenges and other information, keep an eye on [RipVIS website](https://ripvis.ai) or write to andrei.dumitriu@uni-wuerzburg.de .
+
+## Table of Contents
+1. [Short Description](#short-description)
+1. [Links of Interest](#links-of-interest)
+1. [Structure](#structure)
+1. [Splits](#splits)
+1. [Citations](#citations)
+1. [Contributing](#contributing)
+1. [Licensing](#licensing)
+1. [Workshops and Challenges](#workshops-and-challenges)
+1. [Known Limitations](#known-limitations)
+1. [Future Updates](#future-updates)
+1. [Current Version](#current-version)
+
+## Short description
+RipVIS dataset was introduced with [RipVIS: Rip Currents Video Instance Segmentation Benchmark for Beach Monitoring and Safety](https://arxiv.org/abs/2504.01128) paper, accepted at [CVPR 2025](https://cvpr.thecvf.com/Conferences/2025). It is the result of a collaboration of a multi-disciplinary team between [University of Würzburg's](https://www.uni-wuerzburg.de/en/) [Computer Vision Laboratory](https://www.informatik.uni-wuerzburg.de/computervision/) and [University of Bucharest's](https://unibuc.ro/) [Faculty of Mathematics and Computer Science](https://fmi.unibuc.ro/) and [Faculty of Geography](https://fmi.unibuc.ro/).
+
+The dataset consists of 184 videos, out of which 150 videos contain rip currents annotated for instance segmentation. It is authored by:
+- Andrei Dumitriu (andrei.dumitriu@uni-wuerzbuerg.de)
+- Conf. Dr. Florin Tatui
+- Florin Miron
+- Aakash Ralhan
+- Prof. Dr. Radu Ionescu
+- Prof. Dr. Radu Timofte
+
+**[⬆ back to top](#table-of-contents)**
+
+## Links of interest
+- Main contact: andrei.dumitriu@uni-wuerzburg.de
+- Website: https://RipVIS.ai
+- HuggingFace Link: https://huggingface.co/datasets/Irikos/RipVIS/
+- Codabench Evaluation server: coming soon, see ripvis website for updates
+
+**[⬆ back to top](#table-of-contents)**
+
+## Structure
+
+RipVIS folder structure is the following:
+```
+RipVISv1.8.4_instance_segmentation
+    - train
+        -- sampled_images.zip
+            --- images -> folder containing the RipVIS paper sampled images for training split
+            --- addittional_data -> folder containing the extra 2466 images from Dumitriu et al. (CVPRW 2023)
+        -- coco_annotations 
+            --- train.json -> train file with ONLY the data from RipVIS paper
+            --- train_with_additional_data.json -> train file with the data from RipVIS paper and the extra 2466 annotated images from Dumitriu et al. (CVPRW 2023)
+            --- additional_train_data.json -> train file with only the additional data from Dumitriu et al. (CVPRW 2023)
+        -- yolo_annotations.zip -> labels (txt files with train annotations in yolo format)
+        -- videos -> full videos from training split (both with rips and no-rips)
+
+    - val
+        -- sampled_images.zip
+            --- images -> folder containing the RipVIS paper sampled images for validation split
+        -- coco_annotations 
+            --- val.json -> annotations for validation split in coco format
+        -- yolo_annotations.zip -> labels (txt files with validation annotations in yolo format)
+        -- videos -> full videos from validation split (both with rips and no-rips)
+
+    - test
+        -- coco_annotations 
+            --- test_without_annotations.json -> test split in coco format, but with video and frame names from all test split videos, but without annotations. File is provided for structure information when submitting for automatic evaluation.
+        -- videos -> full videos from test split (both with rips and no-rips)
+
+    - compute_coco_ap.py -> script used to evaluate the results and compute AP50 and AP[50:95]
+    - compute_pr_f1_f2.py -> script used to evaluate the results and comptue F1 and F2 scores
+        
+    - RipVISv1.8.4_dataset_info.pdf -> contains information on the dataset (FPS, Duration, Sampling Rate, Resolution, Annotator, Total Sampled Frames and Video Source). Please check "last_update" always when comparing information.
+    - Readme.md (this Readme file)
+```
+
+As exemplifie in the folder structure, we provided the videos and the sampled images alongside their annotations. We also included the "additional data", both images and annotations, which is the training data from our previous paper, [Rip Current Segmentation: A Novel Benchmark and YOLOv8 Baseline Results](https://arxiv.org/abs/2504.02558), presented at CVPRW in 2023. The test data from that paper has been included in the RipVIS data.
+
+For the results in the RipVIS paper, the additional data (2466 images with polygon annotations) has been used in training.
+
+**[⬆ back to top](#table-of-contents)**
+
+
+## Splits
+Due to the amorphous nature of the rip currents, the different locations, video duration and other factors, two of our co-authors, which are rip currents experts, worked in doing a manual train-val-test split. This split structure allows for a reasonable distribution in size, both in video durations and number of annotated frames, ensuring no leakage of data between splits. 
+
+Unfortunately, automatic split of videos (or even worse, of sampled frames) leads to results that do not accuratenly portrait real world results. This can occur due to data leakage (having sampled frames from the same video in multiple splits), uneven distribution of time (splits done on number or videos leading to equal number of videos but of different durations), uneven distribution of orientation or uneven distribution of rip current types. We did our best to minimze all these possible errors in the manual split.
+
+Thus, our recommendation, for obtaining accurate results is to use this split structure. 
+
+<b>The format of the videos and frames is the following:</b>
+1. ```RipVIS-<video_number>``` for videos with rip currents
+1. ```RipVIS-NR-<video_number>```for video without rip currents (NR = No Rips)
+1. ```RipVIS-<video_number>_<frame_number>``` or ```RipVIS-NR-<video_number>_<frame_number>```for frames in video (results should be reported on frames)
+1. video_number is zero-padded to 3 digits. E.g. 4th video is `RipVIS-004.mp4` not `RipVIS-4.mp4`
+1. frame_number is zero-padded to 5 digits. E.g. 28th frames from 4th video is `RipVIS-004_00028.jpg` and NOT `RipVIS-004_28.jpg`. This difference is relevant for the evaluation script.
+
+
+### Train
+RipVIS-
+```['003', '004', '005', '006', '010', '011', '016', '017', '018', '020', '021', '022', '028', '029', '030', '031', '032', '033', '034', '035', '036', '037', '040', '041', '042', '045', '049', '050', '051', '052', '054', '056', '057', '058', '060', '061', '062', '063', '064', '065', '068', '069', '070', '071', '075', '076', '077', '080', '082', '083', '084', '085', '086', '087', '088', '089', '091', '092', '093', '094', '095', '096', '097', '098', '100', '101', '103', '104', '105', '106', '107', '110', '111', '112', '115', '116', '117', '118', '122', '123', '124', '126', '127', '135', '136', '140', '141', '148', '149', '150']```
+
+RipVIS-NR-
+```['002', '003', '006', '007', '008', '009', '010', '011', '012', '015', '016', '017', '018', '025', '027', '028', '029', '030', '031', '032', '033', '034']```
+
+### Validation
+RipVIS-
+```['001', '007', '012', '014', '015', '024', '026', '039', '044', '046', '053', '059', '066', '067', '072', '079', '090', '102', '108', '109', '121', '128', '129', '133', '134', '137', '143', '144', '146', '147']```
+
+RipVIS-NR-
+```['004', '019', '020', '022', '024', '026']```
+
+### Test
+RipVIS-
+```['002', '008', '009', '013', '019', '023', '025', '027', '038', '043', '047', '048', '055', '073', '074', '078', '081', '099', '113', '114', '119', '120', '125', '130', '131', '132', '138', '139', '142', '145']```
+
+RipVIS-NR-
+```['001', '005', '013', '014', '021', '023']```
+
+**[⬆ back to top](#table-of-contents)**
+
+
+## Citations
+Our work is comprised of 3 papers (and many more to come):
+
+1. **RipVIS: Rip Currents Video Instance Segmentation Benchmark for Beach Monitoring and Safety**
+<pre>
+@inproceedings{dumitriu2025ripvis,
+    author    = {Dumitriu, Andrei and Tatui, Florin and Miron, Florin and Ralhan, Aakash and Ionescu, Radu Tudor and Timofte, Radu},
+    title     = {RipVIS: Rip Currents Video Instance Segmentation Benchmark for Beach Monitoring and Safety},
+    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
+    month     = {June},
+    year      = {2025},
+    pages     = {3427--3437}
+}
+</pre>
+
+2. **AIM 2025 Rip Current Segmentation (RipSeg) Challenge**
+<pre>
+@inproceedings{aim2025ripseg,
+    title={{AIM} 2025 Rip Current Segmentation ({RipSeg})} Challenge Report,
+    author={Andrei Dumitriu and Florin Miron and Florin Tatui and Radu Tudor Ionescu and Radu Timofte and Aakash Ralhan and Florin-Alexandru Vasluianu and others},
+    booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops},
+    year={2025}
+}
+</pre>
+
+3. **Rip Current Segmentation: A Novel Benchmark and YOLOv8 Baseline Results**
+<pre>
+@inproceedings{dumitriu2023rip,
+    title="{Rip Current Segmentation: A novel benchmark and YOLOv8 baseline results}",
+    author={Dumitriu, Andrei and Tatui, Florin and Miron, Florin and Ionescu, Radu Tudor and Timofte, Radu},
+    booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
+    pages={1261--1271},
+    year={2023}
+}
+</pre>
+
+**[⬆ back to top](#table-of-contents)**
+
+
+## Contributing
+
+We welcome contributions! Please check out https://RipVIS.ai for more details. Feel free to contact us with any contribution, including suggestions for improving this readme.
+
+### Contributing to the extension of RipVIS Dataset
+We are actively increasing the RipVIS dataset. If you have a video with rip currents, you can send it to us and we will annotate it and include it in the dataset. The video is added under a license decided by you and the video source is credited 100% to you.
+
+
+**[⬆ back to top](#table-of-contents)**
+
+## Licensing
+This dataset is released under the Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0), **with the following additional conditions**:
+
+- Hosting or redistribution of the dataset in its entirety, without explicit written permission, is not permitted.  
+- Redistribution of RipVIS as part of derivative datasets is only permitted if RipVIS constitutes no more than 20% of the resulting dataset. For larger inclusions, prior written permission from the main author is required.  
+
+### In summary:
+1. Non-commercial use only  
+2. Attribution required  
+3. Redistribution conditions (≤ 20% unless permission granted)  
+4. No full-dataset hosting or mirroring without permission  
+
+By downloading or using RipVIS, you agree to these terms.  
+
+A subset of RipVIS that was collected and annotated entirely by the authors is available for commercial licensing. For inquiries regarding commercial use, please contact the main author (andrei.dumitriu@uni-wuerzburg.de).
+
+**[⬆ back to top](#table-of-contents)**
+
+
+## Workshops and Challenges
+1. We organized the [AIM 2025 Rip Current Segmentation (RipSeg) Challenge](https://www.codabench.org/competitions/9109/) challenge at AIM workshop in conjuction with [ICCV2025](https://iccv.thecvf.com/). See the [AIM 2025 Rip Current Segmentation (RipSeg) Challenge Report](https://arxiv.org/abs/2508.13401) on arXiv, which will be published in the ICCVW2025 Proceedings.
+1. Another challenge coming soon, stay tuned.
+
+## Known Limitations
+1. Some of the videos have been annotated with Roboflow. By default, Roboflow sorts files by "Newest". This, however, is rather random when uploading files in bulk. We realised this at some point and sorted the files by filename, leading to annotations in the frame order. However, for the videos annotated with "Newest", the annotations can be quite jittery when overlaied in a video. This does not affect the quantitative results in any observable way and is strictly a visual issue. 
+1. Video sampling rate varies greatly, depending on video duration, movement and annotator disponibility at the specific time. Frames are usually numbered in such a way that they match the exact frame from the video (e.g. RipVIS-015_00005.jpg is the 6th frame from RipVIS-015 video). However, several annotations contain the frames in order, with frame number not matching the actual frame from the video. The actual frame can still be reasonably accurately found by calculating the sampling rate, the total number of frames and the total number of annotated frames. We will fix this in a future patch. 
+Known videos: RipVIS-
+```['001', '003', '004', '006', '007', '011', '012', '014', '016', '017', '018', '020', '021', '022', '024', '033', '034', '035', '036', '037', '039', '040']```.
+1. Some sampled and annotated frames were removed due to invalid annotation format (an error most likely created in transitioning from video to frames to Roboflow and back). While we mitigated this, it is still the case that a a small number of annotated sampled frames might be missing in the published version. 
+
+**[⬆ back to top](#table-of-contents)**
+
+
+## Future Updates
+1. Codabench website for automatic evaluation on the test split (ETA mid October 2025). 
+1. Fixes for known limitation #2.
+1. Manually re-add the #3 in known limitations.
+1. Organizing a new challenge.
+
+**[⬆ back to top](#table-of-contents)**
+
+## Current Version
+Current version of RipVIS is 1.8.4.
+
+Last DATASET update: 25.09.2025 15:40
+Last README update: 27.09.2025 13:00
+
+**[⬆ back to top](#table-of-contents)**
+
+

compute_coco_ap.py

@@ -0,0 +1,97 @@
+"""
+Computes COCO-style evaluation metrics (Average Precision & Average Recall)
+"""
+
+# --- Configure here ---
+GROUND_TRUTH_JSON = "ground_truth.json"
+PREDICTIONS_JSON = "predictions.json"
+IOU_TYPE = "segm"  # "segm", "bbox", or "keypoints"
+OUTPUT_PATH = "results_ap.json"  # set to None to skip saving
+# ----------------------
+
+import json
+from pycocotools.coco import COCO
+from pycocotools.cocoeval import COCOeval
+
+def _load_predictions_for_coco(gt_coco: COCO, predictions_json_path: str):
+    """
+    Loads predictions into COCO's result format.
+
+    Args:
+        gt_coco (COCO): COCO object initialized with ground truth annotations.
+        predictions_json_path (str): Path to predictions JSON file.
+
+    Returns:
+        COCO: A COCO results object that can be passed into COCOeval.
+    """
+    with open(predictions_json_path, "r") as f:
+        data = json.load(f)
+
+    # Normalize predictions into a list of annotations
+    if isinstance(data, list):
+        anns = data
+    elif isinstance(data, dict) and "annotations" in data:
+        anns = data["annotations"]
+    else:
+        raise ValueError("Predictions must be a list or a dict with an 'annotations' key.")
+
+    # Ensure every annotation has a 'score' field (required for COCOeval)
+    for ann in anns:
+        if "score" not in ann:
+            ann["score"] = 1.0  # Assign default score if missing
+
+    # Load predictions into COCO format
+    return gt_coco.loadRes(anns)
+
+
+def compute_ap_map(ground_truth_json: str, predictions_json: str, iou_type: str = "segm"):
+    """
+    Computes COCO-style AP/mAP and AR metrics.
+
+    Args:
+        ground_truth_json (str): Path to COCO-format ground truth file.
+        predictions_json (str): Path to predictions file.
+        iou_type (str): Type of evaluation ("segm", "bbox", or "keypoints").
+
+    Returns:
+        dict: Dictionary containing AP and AR values across IoU thresholds,
+              object sizes, and max detections.
+    """
+    # Load ground truth
+    gt_coco = COCO(ground_truth_json)
+
+    # Load predictions into COCO result format
+    pred_coco = _load_predictions_for_coco(gt_coco, predictions_json)
+
+    # Run COCO evaluation
+    coco_eval = COCOeval(gt_coco, pred_coco, iou_type)
+    coco_eval.evaluate()
+    coco_eval.accumulate()
+    coco_eval.summarize()
+
+    # Collect results from coco_eval.stats (12 values for bbox/segm)
+    stats = coco_eval.stats
+    results = {
+        "AP[0.50:0.95]": float(stats[0]),  # mean AP over IoU thresholds .50:.95
+        "AP@0.50": float(stats[1]),        # AP at IoU=0.50
+        "AP@0.75": float(stats[2]),        # AP at IoU=0.75
+        "AP_small": float(stats[3]),       # AP for small objects
+        "AP_medium": float(stats[4]),      # AP for medium objects
+        "AP_large": float(stats[5]),       # AP for large objects
+        "AR@1": float(stats[6]),           # AR given max 1 detection per image
+        "AR@10": float(stats[7]),          # AR given max 10 detections per image
+        "AR@100": float(stats[8]),         # AR given max 100 detections per image
+        "AR_small": float(stats[9]),       # AR for small objects
+        "AR_medium": float(stats[10]),     # AR for medium objects
+        "AR_large": float(stats[11]),      # AR for large objects
+    }
+    return results
+
+
+if __name__ == "__main__":
+    scores = compute_ap_map(GROUND_TRUTH_JSON, PREDICTIONS_JSON, IOU_TYPE)
+
+    # Optionally save results to JSON
+    if OUTPUT_PATH:
+        with open(OUTPUT_PATH, "w") as f:
+            json.dump(scores, f, indent=2)

compute_pr_f1_f2.py

@@ -0,0 +1,164 @@
+"""
+Computes Precision, Recall, F1, F2. For this script, a confidence threshold variable is required to discard low-score predictions from the model
+"""
+
+
+# --- Configure here ---
+GROUND_TRUTH_JSON = "ground_truth.json"
+PREDICTIONS_JSON = "predictions.json"
+IOU_THRESHOLD = 0.5       # IoU threshold for a TP
+CONFIDENCE_THRESHOLD = 0.1  # ENSURE THAT YOU ADJUST IT ACCORDING TO THE MODEL YOU CHOOSE: predictions with score < this are ignored
+OUTPUT_PATH = "resuts.json"  # set to None to skip saving
+# ----------------------
+
+import json
+import numpy as np
+from pycocotools.coco import COCO
+from pycocotools import mask as maskUtils
+from sklearn.metrics import precision_score, recall_score, f1_score
+
+def _load_and_filter_predictions(predictions_json_path: str, conf_thr: float):
+    """
+    Loads predictions and applies confidence filtering.
+
+    Args:
+        predictions_json_path (str): Path to predictions file.
+        conf_thr (float): Minimum confidence score required to keep a prediction.
+
+    Returns:
+        list: Filtered list of prediction annotations.
+    """
+    with open(predictions_json_path, "r") as f:
+        data = json.load(f)
+
+    # Normalize to list of annotations
+    if isinstance(data, list):
+        anns = data
+    elif isinstance(data, dict) and "annotations" in data:
+        anns = data["annotations"]
+    else:
+        raise ValueError("Predictions must be a list or a dict with an 'annotations' key.")
+
+    # Keep only predictions above confidence threshold
+    filtered = []
+    for ann in anns:
+        score = ann.get("score", 1.0)  # Default score if missing
+        if score >= conf_thr:
+            if "score" not in ann:
+                ann = {**ann, "score": float(score)}
+            filtered.append(ann)
+    return filtered
+
+
+def compute_pr_f1_f2(ground_truth_json: str, predictions_json: str, iou_thr: float, conf_thr: float):
+    """
+    Computes precision, recall, F1, and F2 scores.
+
+    Steps:
+    - Load ground truth annotations.
+    - Load predictions and filter by confidence.
+    - For each image, compute IoU between GT and predicted masks.
+    - Match predictions to GT with highest IoU >= threshold.
+    - Count TP, FP, FN to derive metrics.
+
+    Args:
+        ground_truth_json (str): Path to COCO-format ground truth file.
+        predictions_json (str): Path to predictions file.
+        iou_thr (float): IoU threshold to accept a prediction as True Positive.
+        conf_thr (float): Confidence threshold for filtering predictions.
+
+    Returns:
+        dict: Metrics including precision, recall, F1, F2, and counts of TP, FP, FN.
+    """
+    # Load ground truth
+    gt_coco = COCO(ground_truth_json)
+
+    # Load and filter predictions, then convert to COCO results
+    filtered_preds = _load_and_filter_predictions(predictions_json, conf_thr)
+    pred_coco = gt_coco.loadRes(filtered_preds)
+
+    gt_img_ids = gt_coco.getImgIds()
+    y_true = []
+    y_pred = []
+
+    # Evaluate image by image
+    for img_id in gt_img_ids:
+        gt_ann_ids = gt_coco.getAnnIds(imgIds=img_id)
+        pred_ann_ids = pred_coco.getAnnIds(imgIds=img_id)
+
+        gt_anns = gt_coco.loadAnns(gt_ann_ids)
+        pred_anns = pred_coco.loadAnns(pred_ann_ids)
+
+        # Convert GT and predictions to binary masks
+        gt_masks = [maskUtils.decode(gt_coco.annToRLE(ann)) for ann in gt_anns]
+        pred_masks = [maskUtils.decode(pred_coco.annToRLE(ann)) for ann in pred_anns]
+
+        matched_gt = set()
+        for pred_mask in pred_masks:
+            best_iou = 0.0
+            best_gt_idx = None
+            # Find best IoU match with ground truth masks
+            for i, gt_mask in enumerate(gt_masks):
+                intersection = np.logical_and(gt_mask, pred_mask).sum()
+                union = np.logical_or(gt_mask, pred_mask).sum()
+                iou = (intersection / union) if union > 0 else 0.0
+                if iou > best_iou:
+                    best_iou = iou
+                    best_gt_idx = i
+
+            if best_iou >= iou_thr and best_gt_idx not in matched_gt:
+                # True Positive
+                y_true.append(1)
+                y_pred.append(1)
+                matched_gt.add(best_gt_idx)
+            else:
+                # False Positive
+                y_true.append(0)
+                y_pred.append(1)
+
+        # Unmatched ground truth = False Negatives
+        for i in range(len(gt_masks)):
+            if i not in matched_gt:
+                y_true.append(1)
+                y_pred.append(0)
+
+    # Compute metrics
+    precision = precision_score(y_true, y_pred, zero_division=1)
+    recall = recall_score(y_true, y_pred, zero_division=1)
+    f1 = f1_score(y_true, y_pred, zero_division=1)
+    f2 = (5 * precision * recall) / (4 * precision + recall) if (precision + recall) > 0 else 0.0
+
+    # Count TP, FP, FN 
+    tp = sum(1 for t, p in zip(y_true, y_pred) if t == 1 and p == 1)
+    fp = sum(1 for t, p in zip(y_true, y_pred) if t == 0 and p == 1)
+    fn = sum(1 for t, p in zip(y_true, y_pred) if t == 1 and p == 0)
+
+    results = {
+        "precision": float(precision),
+        "recall": float(recall),
+        "f1": float(f1),
+        "f2": float(f2),
+
+        # Uncomment to print the number of fp, tp, fn as well
+
+        # "tp": int(tp),
+        # "fp": int(fp),
+        # "fn": int(fn),
+    }
+    return results
+
+
+if __name__ == "__main__":
+    # Run evaluation and print results
+    scores = compute_pr_f1_f2(
+        GROUND_TRUTH_JSON,
+        PREDICTIONS_JSON,
+        IOU_THRESHOLD,
+        CONFIDENCE_THRESHOLD,
+    )
+    print(json.dumps(scores, indent=2))
+
+    # Optionally save results to JSON
+    if OUTPUT_PATH:
+        with open(OUTPUT_PATH, "w") as f:
+            json.dump(scores, f, indent=2)

test/videos/RipVIS-002.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6aff802becc36356307f5e2bbd39fb1010b3f0de0b52766f50f2c2469baa3582
+size 12532114

test/videos/RipVIS-008.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab3f3823921af74614ea2cacb36c887e1889d8b73a74d0c160d759f27edbcb78
+size 3122158

test/videos/RipVIS-009.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39a3376176f80d1707f553c07dac7525db27c47c7ebff099a0c98bbc489acd87
+size 6923192

test/videos/RipVIS-013.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b49a0111b9d989f1979594c8ef2590685daff9fab6e9d97aae2177808f13fc46
+size 51324347

test/videos/RipVIS-019.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e88814046b4a59fe24bb4a4266359add3afa22eceaaf6a6caa68fddc5dbbe8ba
+size 21005249

test/videos/RipVIS-023.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e373e78f75735f8dfb74ac5c1d7f561ea5d79c581651bd74a5510a3305a697a3
+size 668941182

test/videos/RipVIS-025.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b48eda789cc0082570b779203cab50393dbf0a6228c994c1316be0a1d68a3d10
+size 94636137

test/videos/RipVIS-027.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6fe57516a473544a30398dbc73b5be8442699625f698c5e47f8b3ed9093e6211
+size 92996724

test/videos/RipVIS-038.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:943b58b0298c3d445390dabde6dd407f3cd0ab6d2e933fdccfeec640ab159056
+size 181232564

test/videos/RipVIS-043.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6612c22b4dfe5ac1f47145e0048211716e8aa23c97a9ff55ce1d062deed110db
+size 519175563

test/videos/RipVIS-047.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1890146ddab390e62f2be09dbbaa5f2318c2023eb42e1c72e4f6e4b2e6502f40
+size 17617249

test/videos/RipVIS-048.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5c4d8d91518a2cff1ed909208d8095aa04147429810200c95416f2bf9a255b81
+size 7607848

test/videos/RipVIS-055.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:22655a85b17db9c86aec3764afa61a2c00a37164af4aa8b72fa931a8599b1d66
+size 7195894

test/videos/RipVIS-073.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:77b62e825c4be334bea010bc9fa2742338437a2b74b2cb24e9ca7c5972a688e1
+size 2284842

test/videos/RipVIS-074.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:317475aaa5b8317dab1fdb7befc4d0b64d1b749a270b3b97388c3295ffb29a23
+size 3326468

test/videos/RipVIS-078.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f7d64394821dcd2c68af696335f467b883fea2876312e922e77d19374a1337a
+size 15464452

test/videos/RipVIS-081.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d2937de7a2bd2ae44a1a95e03355f5bfe16f7688bcada1f86b3ea2cb63a049e
+size 14944545

test/videos/RipVIS-099.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:75e886e28dd5f267e06b54e68eded776592c8e5e04d1d637124bf06b19a9767a
+size 22105916

test/videos/RipVIS-113.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b347565c53dfa659ae34493c53d33bc2d8a5afbefb32d937bcf35cda60e8a7df
+size 238729673

test/videos/RipVIS-114.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c483f5e21a7c6fc9c412d1a34574817eb779295f3c32d67eea7e743375996b0
+size 69784781

test/videos/RipVIS-119.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:63dc6932b0240a759223aaed018e82d0f36fac826913166112118ac0cbbe5220
+size 68900391

test/videos/RipVIS-120.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c5e7dff2383d4c038c970979d2f4b0cd9ecb21e849a7e6d927dc687a7bd94019
+size 2299497

test/videos/RipVIS-125.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f28df7c875b033906fda8cc8949356f735bbc83a5ab112cd5c81e400670eb56
+size 5881187

test/videos/RipVIS-130.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1359f0db646fe3115d8da9836302f69749ebc84d3996508d6bdb45f6f9194e3f
+size 54077722

test/videos/RipVIS-131.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b06882d4a2c9dd267c0602dc316e15d1b36f82b890756d2fdeb4bcf8f79bcf4
+size 35432678

test/videos/RipVIS-132.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8fa95d5d5bc3249be65539be6766ef5a1fc4eda288166f8cece8e21b920d1e37
+size 13101514

test/videos/RipVIS-138.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df6be4853c14212e0738dd0a02baffd912aa42f8ea2698709332562b0f88daf8
+size 4712738

test/videos/RipVIS-139.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4473d95d2ffb5cef8c7827ba172db0a9198afb50eec90d6e7e89d5eabfb3804a
+size 4218334

test/videos/RipVIS-142.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:097e50c6e8f31c7a3d5751d1f1e2185323e07edaad24891d8c26a15f62f7bd7c
+size 54348170

test/videos/RipVIS-145.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:248c99088eb08c1e260b43e4d829d4631c0c302a96f14b2d037f9907a7265de8
+size 138441036

test/videos/RipVIS-NR-001.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:415fd1381ee2c6cf24c566ac0bdd3ede013087aba86ad8c4cb81e26e141d215d
+size 62998066

test/videos/RipVIS-NR-005.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95a07b5987024af832cbb54284b8738c4df319b77de5002b65d33370762361af
+size 72549809

test/videos/RipVIS-NR-013.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d68b14a54ad81f0e121dd4b008d607fbd386688350e527e577a9a136d6a7f23
+size 36636127

test/videos/RipVIS-NR-014.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a3dc36c1c79329d9fcf1833b3528a572d35f1cceb82d6941aca87beb7bb658a
+size 35881166

test/videos/RipVIS-NR-021.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5af06bcd5ee6f75117f9e148ae035758a3667cdaccb7cf1cb8dbb9b57c6e4089
+size 102016372

test/videos/RipVIS-NR-023.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aa1acb7eec9d6a6a33d7e8992df6dbfb4a662ee6cacada73d79dfe4c8f0264cf
+size 98800745

train/coco_annotations/train.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d925b30bf2718169dc221994d855397739b38615555bd7a0685841a069969fec
+size 41419994

train/coco_annotations/train_with_additional_data.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e5b3ab97613607cad28ef148e8d60c20f88bc7505887081b901bcbcf1beba56b
+size 46731720

train/sampled_images.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b86152bfe13b488e77eb45ee1bff0f81cf7cf2e393293985fd8f248aca84a862
+size 6662772952

train/videos/RipVIS-003.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d26c4329520990dbd36afd2c97d5ceaebf4ce7bb91597db28ddc1a36fe53b925
+size 2835712

train/videos/RipVIS-004.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:611de651caf24c002e634360510b6e2762392f6c8c9c50636a93001af6863dc8
+size 18578119

train/videos/RipVIS-005.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f34f119bd59f66c3407288fc5879cd76468e62102d8793fb7039d0d7e11be25c
+size 16414060

train/videos/RipVIS-006.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f3af4f3b499079bf1bd26245eaf9bd8fdfe3039aa80e69cb0585e86c11b394e3
+size 62829316