mppnet_16frames.yaml

CLASS_NAMES: ['Vehicle', 'Pedestrian', 'Cyclist']

DATA_CONFIG:
    _BASE_CONFIG_: cfgs/dataset_configs/waymo_dataset.yaml
    PROCESSED_DATA_TAG: 'waymo_processed_data_v0_5_0'

    SAMPLED_INTERVAL: {
        'train': 1,
        'test': 1
    }
    FILTER_EMPTY_BOXES_FOR_TRAIN: True
    DISABLE_NLZ_FLAG_ON_POINTS: True

    SEQUENCE_CONFIG:
        ENABLED: True
        SAMPLE_OFFSET: [-15,0]
        
    USE_PREDBOX: True
    ROI_BOXES_PATH: {
        'train': '../output/xxxxx/train/result.pkl',  # example: predicted boxes of RPN in training set
        'test': '../output/xxxxx/val/result.pkl',  # example: predicted boxes of RPN in evalulation set
    }

    DATA_AUGMENTOR:
        DISABLE_AUG_LIST: [ 'placeholder' ]
        AUG_CONFIG_LIST:
            -   NAME: random_world_flip
                ALONG_AXIS_LIST: [ 'x', 'y' ]

            -   NAME: random_world_rotation
                WORLD_ROT_ANGLE: [ -0.78539816, 0.78539816 ]

            -   NAME: random_world_scaling
                WORLD_SCALE_RANGE: [ 0.95, 1.05 ]

    DATA_PROCESSOR:
        -   NAME: mask_points_and_boxes_outside_range
            REMOVE_OUTSIDE_BOXES: True

        -   NAME: shuffle_points
            SHUFFLE_ENABLED: {
                'train': True,
                'test': True
            }


    POINT_FEATURE_ENCODING: {
        encoding_type: absolute_coordinates_encoding,
        used_feature_list: ['x', 'y', 'z', 'intensity', 'elongation', 'time'],
        src_feature_list: ['x', 'y', 'z', 'intensity', 'elongation', 'time'],       
        }


MODEL:
    NAME: MPPNet

    ROI_HEAD:
        NAME: MPPNetHead
        TRANS_INPUT: 64
        CLASS_AGNOSTIC: True
        USE_BOX_ENCODING:
            ENABLED: True 
        AVG_STAGE1_SCORE: True
        USE_TRAJ_EMPTY_MASK: True
        USE_AUX_LOSS: True
        IOU_WEIGHT: [0.5,0.4]

        ROI_GRID_POOL:
            GRID_SIZE: 4
            MLPS: [[64,64]]
            POOL_RADIUS: [0.8]
            NSAMPLE: [16]
            POOL_METHOD: max_pool

        Transformer:
            num_lidar_points: 128
            num_proxy_points: 64  # GRID_SIZE*GRID_SIZE*GRID_SIZE
            pos_hidden_dim: 64
            enc_layers: 3
            dim_feedforward: 512
            hidden_dim: 64 #equal to ROI_HEAD.TRANS_INPUT
            dropout: 0.1
            nheads: 4
            pre_norm: False
            num_frames: 16
            num_groups: 4
            sequence_stride: 4
            use_grid_pos:
                enabled: True
                init_type: index
            use_mlp_mixer:
                enabled: True
                hidden_dim: 16

        TARGET_CONFIG:
            BOX_CODER: ResidualCoder
            ROI_PER_IMAGE: 96
            FG_RATIO: 0.5
            REG_AUG_METHOD: single
            ROI_FG_AUG_TIMES: 10
            RATIO: 0.2
            USE_ROI_AUG: True
            USE_TRAJ_AUG:
                ENABLED: True
                THRESHOD: 0.8
            SAMPLE_ROI_BY_EACH_CLASS: True
            CLS_SCORE_TYPE: roi_iou

            CLS_FG_THRESH: 0.75
            CLS_BG_THRESH: 0.25
            CLS_BG_THRESH_LO: 0.1
            HARD_BG_RATIO: 0.8

            REG_FG_THRESH: 0.55

        LOSS_CONFIG:
            CLS_LOSS: BinaryCrossEntropy
            REG_LOSS: smooth-l1
            CORNER_LOSS_REGULARIZATION: True
            LOSS_WEIGHTS: {
                'rcnn_cls_weight': 1.0,
                'rcnn_reg_weight': 1.0,
                'rcnn_corner_weight': 2.0,
                'traj_reg_weight': [2.0, 2.0, 2.0],
                'code_weights': [1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]
            }

    POST_PROCESSING:
        RECALL_THRESH_LIST: [0.3, 0.5, 0.7]
        SCORE_THRESH: 0.1
        OUTPUT_RAW_SCORE: False
        SAVE_BBOX: False
        EVAL_METRIC: waymo
        NOT_APPLY_NMS_FOR_VEL: True

        NMS_CONFIG:
            MULTI_CLASSES_NMS: False
            NMS_TYPE: nms_gpu
            NMS_THRESH: 0.7
            NMS_PRE_MAXSIZE: 4096
            NMS_POST_MAXSIZE: 500

OPTIMIZATION:
    BATCH_SIZE_PER_GPU: 2
    NUM_EPOCHS: 6

    OPTIMIZER: adam_onecycle
    LR: 0.003
    WEIGHT_DECAY: 0.01
    MOMENTUM: 0.9

    MOMS: [0.95, 0.85]
    PCT_START: 0.4
    DIV_FACTOR: 10
    DECAY_STEP_LIST: [35, 45]
    LR_DECAY: 0.1
    LR_CLIP: 0.0000001

    LR_WARMUP: False
    WARMUP_EPOCH: 1

    GRAD_NORM_CLIP: 10