fix some typos

README.md

@@ -109,8 +109,12 @@ import mmscan
 # (1) The dataset tool
 import mmscan.MMScan as MMScan_dataset
 
-# (2) The evaluator tool ('VG', 'QA', 'GPT')
-import mmscan.{}_Evaluator as MMScan_{}_evaluator
+# (2) The evaluator tool ('VisualGroundingEvaluator', 'QuestionAnsweringEvaluator', 'GPTEvaluator')
+import mmscan.VisualGroundingEvaluator as MMScan_VG_evaluator
+
+import mmscan.QuestionAnsweringEvaluator as MMScan_QA_evaluator
+
+import mmscan.GPTEvaluator as MMScan_GPT_evaluator
 ```
 
 ### MMScan Dataset
@@ -124,7 +128,7 @@ Initialize the dataset for a specific task with:
 ```bash
 my_dataset = MMScan_dataset(split='train', task="MMScan-QA", ratio=1.0)
 # Access a specific sample
-print(my_dataset[100])
+print(my_dataset[index])
 ```
 
 #### Data Access

data_preparation/README.md

@@ -9,10 +9,10 @@ Detailed steps are shown as follows.
 
 3. Download Matterport3D data [HERE](https://github.com/niessner/Matterport). Link or move the folder to this level of directory.
 
-4. Organize the file structure. You are recommanded to create a soft link to the raw data folder under `mmsan_data/embodiedscan-split/data`.
+4. Organize the file structure. You are recommanded to create a soft link to the raw data folder under `mmscan_data/embodiedscan-split/data`.
 
    ```
-   mmsan_data/embodiedscan-split/data/
+   mmscan_data/embodiedscan-split/data/
    ├── scannet/
    │   ├── scans
    │   │   ├── <scene_id>
@@ -25,10 +25,10 @@ Detailed steps are shown as follows.
    │   ├── ...
    ```
 
-   Additionally, create a `process_pcd` folder under `mmsan_data/embodiedscan-split` to store the results. Similarly, we recommend using a symbolic link, as the total file size might be a little large (approximately 21GB)
+   Additionally, create a `process_pcd` folder under `mmscan_data/embodiedscan-split` to store the results. Similarly, we recommend using a symbolic link, as the total file size might be a little large (approximately 21GB)
 
    PS: If you have followed the embodiedscan tutorial to organize the data, you can skip these steps and link or copy the `data` folder to
-   `mmsan_data/embodiedscan-split`.
+   `mmscan_data/embodiedscan-split`.
 
    After all the raw data is organized, the directory structure should be as below:
 

data_preparation/process_all_scan.py

@@ -1,6 +1,7 @@
 import json
 import os
 from argparse import ArgumentParser
+from functools import wraps
 from typing import Dict, Tuple
 
 import mmengine
@@ -11,13 +12,22 @@
 from utils.data_utils import read_annotation_pickle
 from utils.mp3d_process import process_mp3d
 from utils.pcd_utils import is_inside_box
-from utils.proc_utils import mmengine_track_func
 from utils.scannet_process import process_scannet
-from utils.trscan_process import process_trscan
+from utils.trscan_process import process_3rscan
 
 es_anno = {}
 
 
+def mmengine_track_func(func):
+
+    @wraps(func)
+    def wrapped_func(args):
+        result = func(*args)
+        return result
+
+    return wrapped_func
+
+
 def create_scene_pcd(es_anno: dict,
                      pcd_result: Tuple[np.ndarray, np.ndarray, np.ndarray]) \
                         -> Tuple[np.ndarray, np.ndarray,
@@ -135,7 +145,7 @@ def process_one_scan(
                 return
             pcd_info = create_scene_pcd(
                 es_anno['3rscan/' + scan_id],
-                process_trscan(scan_id, trscan_root, trscan_matrix),
+                process_3rscan(scan_id, trscan_root, trscan_matrix),
             )
 
         save_path = f'{save_root}/{scan_id}.pth'

data_preparation/utils/mp3d_process.py

@@ -7,7 +7,7 @@ def process_mp3d(new_scan_id: str, data_root: str,
     """Process matterport3d data.
 
     Args:
-        scan_id (str): ID of the matterport3d scan.
+        new_scan_id (str): processed ID of the matterport3d scan.
         data_root (str): Root directory of the matterport3d dataset.
         axis_align_matrix_dict (dict): Dict of axis alignment matrices
             for each scan.

data_preparation/utils/pcd_utils.py

@@ -38,6 +38,7 @@ def is_inside_box(points: np.ndarray, center: np.ndarray, size: np.ndarray,
         size(np.ndarray): size of the box, numpy array of shape (3, ).
         rotation_mat(np.ndarray): rotation matrix of the box,
             numpy array of shape (3, 3).
+
     Returns:
         np.ndarray: Boolean array of shape (n, ) indicating if each point
             is inside the box.
@@ -121,6 +122,7 @@ def euler_to_matrix_np(euler):
 
     Args:
         euler (np.ndarray) : (..., 3)
+
     Returns:
         np.ndarray : (..., 3, 3)
     """

data_preparation/utils/trscan_process.py

@@ -5,7 +5,7 @@
 from pytorch3d.io import load_obj
 
 
-def process_trscan(scan_id: str, data_root: str, axis_align: dict):
+def process_3rscan(scan_id: str, data_root: str, axis_align: dict):
     """Process 3rscan data.
 
     Args:

mmscan/__init__.py

@@ -3,14 +3,14 @@
 
 print('MMScan module loaded')
 try:
-    from mmscan.evaluator.vg_evaluation import VG_Evaluator
+    from mmscan.evaluator.vg_evaluation import VisualGroundingEvaluator
 except:
     pass
 try:
-    from mmscan.evaluator.qa_evaluation import QA_Evaluator
+    from mmscan.evaluator.qa_evaluation import QuestionAnsweringEvaluator
 except:
     pass
 try:
-    from mmscan.evaluator.gpt_evaluation import GPT_Evaluator
+    from mmscan.evaluator.gpt_evaluation import GPTEvaluator
 except:
     pass

mmscan/evaluator/gpt_evaluation.py

@@ -9,7 +9,7 @@
 from mmscan.utils.lang_utils import qa_metric_map, qa_prompt_define
 
 
-class GPT_Evaluator:
+class GPTEvaluator:
     """GPT metric, we set this for QA and Caption tasks.
 
     Args:
@@ -55,6 +55,7 @@ def normal_query(self,
             user_content_grounps (list[str]) :
                 The user content inputted into GPT.
             max_tokens (int) : Max tokens. Defaults to 1000.
+
         Returns:
             dict : The json-format result.
         """
@@ -136,6 +137,7 @@ def qa_collection(self, num_threads: int, tmp_path: str) -> dict:
                 The number of threads used to evaluate the samples.
             tmp_path (str) :
                 The path to store the tmp-stored json files.
+
         Returns:
             dict : The evaluation result.
         """
@@ -181,6 +183,7 @@ def load_and_eval(self,
             num_threads (int) : The number of the threadings.
                 Defaults to 1.
             tmp_path (str) : The temporay path to store the json files.
+
         Returns:
             dict : The evaluation result.
         """

mmscan/evaluator/qa_evaluation.py

@@ -8,7 +8,7 @@
 from mmscan.utils.lang_utils import special_token_filter
 
 
-class QA_Evaluator:
+class QuestionAnsweringEvaluator:
     """Tradition metrics for QA and Caption evaluation , consists the
     implements of.
 
@@ -81,6 +81,7 @@ def update(self, batch_input: List[dict]) -> dict:
         Args:
             batch_input (list[dict]):
                 Batch of the raw original input.
+
         Returns:
             Dict: {"EM":EM metric for this batch,
                 "refined_EM":Refined EM metric for this batch}

mmscan/evaluator/vg_evaluation.py

@@ -11,7 +11,7 @@
 from mmscan.utils.box_utils import index_box, to_9dof_box
 
 
-class VG_Evaluator:
+class VisualGroundingEvaluator:
     """Evaluator for MMScan Visual Grounding benchmark. The evaluation metric
     includes "AP","AP_C","AR","gTop-k".
 

mmscan/mmscan.py

@@ -216,6 +216,7 @@ def __getitem__(self, index_: int) -> dict:
 
         Args:
             index_ (int): The index.
+
         Returns:
             dict: The sample item corresponding to the index.
         """
@@ -281,6 +282,7 @@ def get_possess(self, table_name: str, scan_idx: str):
         Args:
             table_name (str): The ype of the expected data.
             scan_idx (str): The scan id to get the data.
+
         Returns:
             The data corresponding to the table_name and scan_idx.
         """
@@ -330,6 +332,7 @@ def __filter_lang_anno__(self, samples: List[dict]) -> List[dict]:
 
         Args:
             samples (list[dict]): The samples.
+
         Returns:
             list[dict] : The filtered results.
         """
@@ -348,6 +351,7 @@ def __check_lang_anno__(self, sample: dict) -> bool:
 
         Args:
             sample (dict): The item from the samples.
+
         Returns:
             bool : Whether the item is valid or not.
         """
@@ -375,6 +379,7 @@ def __load_base_anno__(self, pkl_path: str) -> dict:
 
         Args:
             pkl_path (str): The path of the pkl.
+
         Returns:
             dict : The embodiedscan annotations of scans.
             (with scan_idx as keys)
@@ -388,6 +393,7 @@ def __process_pcd_info__(self, scan_idx: str) -> dict:
 
         Args:
             scan_idx (str): ID of the scan.
+
         Returns:
             dict : The corresponding scan information.
         """
@@ -423,6 +429,7 @@ def __process_box_info__(self, scan_idx: str) -> dict:
 
         Args:
             scan_idx (str): ID of the scan.
+
         Returns:
             dict : The corresponding bounding boxes information.
         """
@@ -447,6 +454,7 @@ def __process_img_info__(self, scan_idx: str) -> List[dict]:
 
         Args:
             scan_idx (str): ID of the scan.
+
         Returns:
             list[dict] :The corresponding bounding boxes information
                 for each camera.
@@ -487,6 +495,7 @@ def down_9dof_to_6dof(
                 the point clouds
             box_9DOF(np.ndarray / Tensor):
                 the 9DOF bounding box
+
         Returns:
             np.ndarray :
                 The transformed 6DOF bounding box.
@@ -501,6 +510,7 @@ def __downsample_annos__(self, annos: List[dict],
         Args:
             annos (list[dict]): The original annotations.
             ratio (float): The ratio to downsample.
+
         Returns:
             list[dict] : The result.
         """

mmscan/utils/lang_utils.py

@@ -10,6 +10,7 @@ def clean_answer(data):
 
     Args:
         data (str): the raw sentence.
+
     Returns:
         data (str): the processed sentence.
     """

mmscan/utils/task_utils.py

@@ -10,6 +10,7 @@ def anno_token_flatten(samples: List[dict], keep_only_one: bool = True):
         keep_only_one (bool):
             Whether to keep only one positive token for each target.
             Defaults to True.
+
     Returns:
         List[dict] : The token-flattened samples.
     """

models/EmbodiedScan/embodiedscan/eval/metrics/grounding_metric_mod.py

@@ -13,7 +13,7 @@
 from terminaltables import AsciiTable
 from tqdm import tqdm
 
-from mmscan import VG_Evaluator
+from mmscan import VisualGroundingEvaluator
 
 
 def abbr(sub_class):
@@ -62,7 +62,7 @@ def __init__(self,
         self.prefix = prefix
         self.format_only = format_only
         self.result_dir = result_dir
-        self.mmscan_eval = VG_Evaluator(True)
+        self.mmscan_eval = VisualGroundingEvaluator(True)
 
     def to_mmscan_form(self, det_annos, gt_annos):
         batch_input = []

models/LEO/evaluator/GPT_eval.py

@@ -2,7 +2,7 @@
 from argparse import ArgumentParser
 from pathlib import Path
 
-from mmscan import GPT_Evaluator
+from mmscan import GPTEvaluator
 
 if __name__ == '__main__':
     parser = ArgumentParser()
@@ -15,7 +15,7 @@
 
     leo_file_path = args.file
 
-    evaluator = GPT_Evaluator(eval_size =args.eval_size,\
+    evaluator = GPTEvaluator(eval_size =args.eval_size,\
         API_key=args.api_key)
 
     with open(leo_file_path, 'r') as f:

models/LEO/evaluator/mmscan_eval.py

@@ -1,7 +1,7 @@
 import json
 from pathlib import Path
 
-from mmscan import QA_Evaluator
+from mmscan import QuestionAnsweringEvaluator
 
 model_config = {'simcse': '', 'sbert': ''}
 
@@ -12,7 +12,7 @@
 class MMScanEvaluator():
 
     def __init__(self, cfg, task_name):
-        self.evaluator = QA_Evaluator(model_config)
+        self.evaluator = QuestionAnsweringEvaluator(model_config)
         self.task_name = task_name
         self.target_metric = 'refined_EM'
         self.best_result = 0.0

models/LL3DA/data/scannet/meta_data/render_option.json

@@ -37,4 +37,4 @@
 	"show_coordinate_frame" : false,
 	"version_major" : 1,
 	"version_minor" : 0
-}
+}

models/LL3DA/data/scannet_category_map.json

@@ -606,4 +606,4 @@
     "food bag": "others",
     "clothes hangers": "others",
     "starbucks cup": "others"
-}
+}

models/LL3DA/eval_utils/evaluate_gpt.py

@@ -1,7 +1,7 @@
 import json
 from argparse import ArgumentParser
 
-from mmscan import GPT_Evaluator
+from mmscan import GPTEvaluator
 
 
 def parse_form(results):
@@ -28,7 +28,7 @@ def parse_form(results):
 
     ll3da_file_path = args.file
 
-    evaluator = GPT_Evaluator(eval_size =args.eval_size,\
+    evaluator = GPTEvaluator(eval_size =args.eval_size,\
         API_key=args.api_key)
 
     with open(ll3da_file_path, 'r') as f: