--- title: panoptic-quality tags: - evaluate - metric description: PanopticQuality score sdk: gradio sdk_version: 3.19.1 app_file: app.py pinned: false emoji: 🖼️ --- # SEA-AI/PanopticQuality This hugging face metric uses `seametrics.segmentation.PanopticQuality` under the hood to compute a panoptic quality score. It is a wrapper class for the torchmetrics class [`torchmetrics.detection.PanopticQuality`](https://lightning.ai/docs/torchmetrics/stable/detection/panoptic_quality.html). ## Getting Started To get started with PanopticQuality, make sure you have the necessary dependencies installed. This metric relies on the `evaluate`, `seametrics` and `seametrics[segmentation]`libraries for metric calculation and integration with FiftyOne datasets. ### Basic Usage ```python >>> import evaluate >>> from seametrics.payload.processor import PayloadProcessor >>> MODEL_FIELD = ["maskformer-27k-100ep"] >>> payload = PayloadProcessor("SAILING_PANOPTIC_DATASET_QA", >>> gt_field="ground_truth_det", >>> models=MODEL_FIELD, >>> sequence_list=["Trip_55_Seq_2", "Trip_197_Seq_1", "Trip_197_Seq_68"], >>> excluded_classes=[""]).payload >>> module = evaluate.load("SEA-AI/PanopticQuality") >>> module.add_payload(payload, model_name=MODEL_FIELD[0]) >>> module.compute() 100%|██████████| 3/3 [00:03<00:00, 1.30s/it] Added data ... Start computing ... Finished! {'scores': {'MOTORBOAT': [0.18632257426639526, 0.698709617058436, 0.2666666805744171], 'FAR_AWAY_OBJECT': [0.0, 0.0, 0.0], 'SAILING_BOAT_WITH_CLOSED_SAILS': [0.0, 0.0, 0.0], 'SHIP': [0.3621737026917471, 0.684105846616957, 0.529411792755127], 'WATERCRAFT': [0.0, 0.0, 0.0], 'SPHERICAL_BUOY': [0.0, 0.0, 0.0], 'FLOTSAM': [0.0, 0.0, 0.0], 'SAILING_BOAT_WITH_OPEN_SAILS': [0.0, 0.0, 0.0], 'CONTAINER': [0.0, 0.0, 0.0], 'PILLAR_BUOY': [0.0, 0.0, 0.0], 'AERIAL_ANIMAL': [0.0, 0.0, 0.0], 'HUMAN_IN_WATER': [0.0, 0.0, 0.0], 'WOODEN_LOG': [0.0, 0.0, 0.0], 'MARITIME_ANIMAL': [0.0, 0.0, 0.0], 'WATER': [0.9397601008415222, 0.9397601008415222, 1.0], 'SKY': [0.9674496332804362, 0.9674496332804362, 1.0], 'LAND': [0.30757412078761204, 0.8304501533508301, 0.37037035822868347], 'CONSTRUCTION': [0.0, 0.0, 0.0], 'OWN_BOAT': [0.0, 0.0, 0.0], 'ALL': [0.14543579641409013, 0.21686712374464112, 0.16665520166095935]}, 'numbers': {'MOTORBOAT': [6, 15, 18, 4.1922577023506165], 'FAR_AWAY_OBJECT': [0, 8, 9, 0.0], 'SAILING_BOAT_WITH_CLOSED_SAILS': [0, 2, 0, 0.0], 'SHIP': [9, 1, 15, 6.156952619552612], 'WATERCRAFT': [0, 9, 12, 0.0], 'SPHERICAL_BUOY': [0, 4, 22, 0.0], 'FLOTSAM': [0, 0, 1, 0.0], 'SAILING_BOAT_WITH_OPEN_SAILS': [0, 6, 0, 0.0], 'CONTAINER': [0, 0, 0, 0.0], 'PILLAR_BUOY': [0, 0, 9, 0.0], 'AERIAL_ANIMAL': [0, 0, 0, 0.0], 'HUMAN_IN_WATER': [0, 0, 0, 0.0], 'WOODEN_LOG': [0, 0, 0, 0.0], 'MARITIME_ANIMAL': [0, 0, 0, 0.0], 'WATER': [15, 0, 0, 14.096401512622833], 'SKY': [15, 0, 0, 14.511744499206543], 'LAND': [5, 9, 8, 4.15225076675415], 'CONSTRUCTION': [0, 0, 0, 0.0], 'OWN_BOAT': [0, 0, 8, 0.0], 'ALL': [50, 54, 102, 43.109607100486755]}} ``` ## Metric Settings The metric takes four optional input parameters: __label2id__, __stuff__, __per_class__ and __split_sq_rq__. * `label2id: Dict[str, int]`: this dictionary is used to map string labels to an integer representation. if not provided a default setting will be used: `{'WATER': 0, 'SKY': 1, 'LAND': 2, 'MOTORBOAT': 3, 'FAR_AWAY_OBJECT': 4, 'SAILING_BOAT_WITH_CLOSED_SAILS': 5, 'SHIP': 6, 'WATERCRAFT': 7, 'SPHERICAL_BUOY': 8, 'CONSTRUCTION': 9, 'FLOTSAM': 10, 'SAILING_BOAT_WITH_OPEN_SAILS': 11, 'CONTAINER': 12, 'PILLAR_BUOY': 13, 'AERIAL_ANIMAL': 14, 'HUMAN_IN_WATER': 15, 'OWN_BOAT': 16, 'WOODEN_LOG': 17, 'MARITIME_ANIMAL': 18} ` * `stuff: List[str]`: this list holds all string labels that belong to stuff. if not provided a default setting will be used: ` ["WATER", "SKY", "LAND", "CONSTRUCTION", "ICE", "OWN_BOAT"]` * `per_class: bool = True`: By default, the results are split up per class. Setting this to False will aggregate the results: - average the "scores" - sum up the "numbers" * `split_sq_rq: bool = True`: By default, the PQ-score is returned in three parts: the PQ score itself, and split into the segmentation quality (SQ) and recognition quality (RQ) part. Setting this to False will aggregate return the PQ score only (PQ=RQ*SQ). ## Output Values A dictionary containing the following keys: * __scores__: This is a dictionary, that contains a key for each label, if `per_class == True`. Otherwise it only contains the key __all__. For each key, it contains a list that holds the scores in the following order: PQ, SQ and RQ. If `split_sq_rq == False`, the list consists of PQ only. * __numbers__: This is a dictionary, that contains a key for each label, if `per_class == True`. Otherwise it only contains the key __all__. For each key, it contains a list that consists of four elements: TP, FP, FN and IOU: - __TP__: number of true positive predictions - __FP__: number of false positive predictions - __FN__: number of false negative predictions - __IOU__: sum of IOU of all TP predictions with ground truth With all these values, it is possible to calculate the final scores. ## Further References - **seametrics Library**: Explore the [seametrics GitHub repository](https://github.com/SEA-AI/seametrics/tree/main) for more details on the underlying library. - **Torchmetrics**: https://lightning.ai/docs/torchmetrics/stable/detection/panoptic_quality.html - **Understanding Metrics**: The Panoptic Segmentation task, as well as Panoptic Quality as the evaluation metric, were introduced [in this paper](https://arxiv.org/pdf/1801.00868.pdf). ## Contribution Your contributions are welcome! If you'd like to improve SEA-AI/PanopticQuality or add new features, please feel free to fork the repository, make your changes, and submit a pull request.