Tennis table dataset (#29)

* feat: added datasets via DVC

feat: added new parser from binary .out file to raw png

* fix: minor codestyle

* feat: added dataset for 2ms exposure

* feat: added sorted frames json files for orange dark 2ms frames

* feat: sorted dataset orange 2ms

* feat: stats for frames in dataset

* fix: replaced annotation

* feat: segmentation masks for dataset

* feat: added recorded parameters to dataset

* fix: yaml codestyle

* feat: annotated datasets

---------

Co-authored-by: Daniel Paleyev <danny.paleyev@gmail.com>
Co-authored-by: Denis Bakin <denis.bakin@gmail.com>

datasets/README.md

@@ -0,0 +1,43 @@
+# Information about datasets
+
+## TableMulticolorRecord_19_04
+
+|  |  |
+| --- | --- |
+| Date | 19.04.2024 |
+| Location | Table tennis club HSE |
+| Setup | 2 cameras across the one side of the table, ~90 degress axis intersection |
+| Calibration available | No |
+| Annotation | \<ball-color>\_<light/dark>\_\<exposure> |
+
+### Available sets
+- 6ms exposure
+    - With additional light (from camera ID=2)
+        - green
+        - orange
+        - white
+    - No additional light
+        - green
+        - orange
+        - white
+
+
+## TableOrange2msRecord_22_04
+
+|  |  |
+| --- | --- |
+| Date | 22.04.2024 |
+| Location | Table tennis club HSE |
+| Setup | 2 cameras across the one side of the table, ~90 degress axis intersection |
+| Calibration available | Yes (intrinsics + stereo) |
+| Annotation | \<ball-color>\_<light/dark>\_\<exposure> |
+
+### Available sets
+- 2ms exposure
+    - With additional light (from camera ID=2)
+        - orange
+    - No additional light
+        - orange
+- calibration boards
+    - aruco board (GridBoard, 5x7 cells, 60mm marker, 30mm gap, DICT_5X5_250)
+    - charuco board (CharucoBoard, 7x10 cells, 60mm cell, 40mm marker, DICT_5X5_250)

datasets/TableMulticolorRecord_19_04/.gitignore

@@ -0,0 +1,6 @@
+/green_dark_6ms
+/green_light_6ms
+/orange_dark_6ms
+/orange_light_6ms
+/white_light_6ms
+/white_dark_6ms

datasets/TableMulticolorRecord_19_04/green_dark_6ms.dvc

@@ -0,0 +1,6 @@
+outs:
+- md5: a931100ed10f3117ccde87999550b6c9.dir
+  size: 8261648570
+  nfiles: 6000
+  hash: md5
+  path: green_dark_6ms

datasets/TableMulticolorRecord_19_04/green_light_6ms.dvc

@@ -0,0 +1,6 @@
+outs:
+- md5: 90b646460cabb3129a4460637c474d15.dir
+  size: 11967419975
+  nfiles: 8000
+  hash: md5
+  path: green_light_6ms

datasets/TableMulticolorRecord_19_04/orange_dark_6ms.dvc

@@ -0,0 +1,6 @@
+outs:
+- md5: 942ea6de714878c631b4b3cdf1ffff26.dir
+  size: 8262419969
+  nfiles: 6000
+  hash: md5
+  path: orange_dark_6ms

datasets/TableMulticolorRecord_19_04/orange_light_6ms.dvc

@@ -0,0 +1,6 @@
+outs:
+- md5: 4067b7966c1eff87056fe72ab50e4bb2.dir
+  size: 11803559470
+  nfiles: 8000
+  hash: md5
+  path: orange_light_6ms

datasets/TableMulticolorRecord_19_04/white_dark_6ms.dvc

@@ -0,0 +1,6 @@
+outs:
+- md5: 7abcfb63c142dc9b88ba553746cdc404.dir
+  size: 5512126747
+  nfiles: 4000
+  hash: md5
+  path: white_dark_6ms

datasets/TableMulticolorRecord_19_04/white_light_6ms.dvc

@@ -0,0 +1,6 @@
+outs:
+- md5: 12fa64fd477079cbde046101bee39dcb.dir
+  size: 5985412843
+  nfiles: 4000
+  hash: md5
+  path: white_light_6ms

datasets/TableOrange2msRecord_22_04/.gitignore

@@ -0,0 +1,3 @@
+/orange_dark_2ms
+/orange_light_2ms
+/calibration_boards

datasets/TableOrange2msRecord_22_04/README.md

@@ -0,0 +1,25 @@
+# Dataset info
+
+## Frame statistics
+
+|                      |   no ball |   ball |   too many balls |
+|:---------------------|----------:|-------:|-----------------:|
+| orange_light_2ms_1_1 |       554 |    446 |                0 |
+| orange_light_2ms_1_2 |       550 |    450 |                0 |
+| orange_light_2ms_2_1 |       493 |    339 |              168 |
+| orange_light_2ms_2_2 |       200 |    578 |              222 |
+| orange_light_2ms_3_1 |       515 |    482 |                3 |
+| orange_light_2ms_3_2 |       571 |    429 |                0 |
+| orange_light_2ms_4_1 |       459 |    529 |               12 |
+| orange_light_2ms_4_2 |       488 |    467 |               45 |
+
+|                     |   no ball |   ball |
+|:--------------------|----------:|-------:|
+| orange_dark_2ms_1_1 |       505 |    495 |
+| orange_dark_2ms_1_2 |       489 |    511 |
+| orange_dark_2ms_2_1 |       421 |    579 |
+| orange_dark_2ms_2_2 |       333 |    667 |
+| orange_dark_2ms_3_1 |       483 |    517 |
+| orange_dark_2ms_3_2 |       382 |    618 |
+| orange_dark_2ms_4_1 |       562 |    438 |
+| orange_dark_2ms_4_2 |       523 |    477 |

datasets/TableOrange2msRecord_22_04/calibration_boards.dvc

@@ -0,0 +1,6 @@
+outs:
+- md5: c1220b972df0c1240333c5a026c4625e.dir
+  size: 14854427229
+  nfiles: 8000
+  hash: md5
+  path: calibration_boards

datasets/TableOrange2msRecord_22_04/camera_params.yaml

@@ -0,0 +1,15 @@
+parameters:
+  1:
+    image_size: [1920, 1200]
+    camera_matrix: [672.2824725267757, 0, 984.0472159818853, 0, 672.6886411532304, 602.96669930345, 0, 0, 1]
+    distortion_coefs: [-0.09715103386082896, 0.06788948036532018,
+                      -0.0007157453506997161, 0.0003048354358359307, -0.003636308978789861]
+    rotation: [0, 0, 0]
+    translation: [0, 0, 0]
+  2:
+    image_size: [1920, 1200]
+    camera_matrix: [685.7143789189881, 0, 991.0247637161314, 0, 686.3020333004097, 601.2442243349392, 0, 0, 1]
+    distortion_coefs: [-0.09781628655937251, 0.07153618281495966,
+                       -0.001066517414175782, 0.0004679942401339674, -0.003645360450147547]
+    rotation: [0.01175419895518242, 2.170836441913732, 2.19333242876324]
+    translation: [-0.06677747450343367, -1.174973690204363, 1.140354306665756]

datasets/TableOrange2msRecord_22_04/orange_dark_2ms.dvc

@@ -0,0 +1,6 @@
+outs:
+- md5: c02049c918851de465ae6aa8b2699fe3.dir
+  size: 12110150634
+  nfiles: 10251
+  hash: md5
+  path: orange_dark_2ms

datasets/TableOrange2msRecord_22_04/orange_light_2ms.dvc

@@ -0,0 +1,6 @@
+outs:
+- md5: 80d3d47cb9c6043bf9fcf485373acfc4.dir
+  size: 12751192086
+  nfiles: 9814
+  hash: md5
+  path: orange_light_2ms

packages/camera/scripts/binary_parser.py

@@ -0,0 +1,56 @@
+import argparse
+import os
+from concurrent.futures import ThreadPoolExecutor
+
+import cv2
+import numpy as np
+from tqdm import tqdm
+
+
+def save_image(i, img_data, output_folder, width, height):
+    img = np.frombuffer(img_data, dtype=np.uint8).reshape((height, width))
+    cv2.imwrite(os.path.join(output_folder, f"image_{str(i).rjust(4, '0')}.png"), img)
+
+
+def read_and_convert(file_path, output_folder, width=1920, height=1200):
+    if not os.path.exists(output_folder):
+        os.makedirs(output_folder)
+
+    num_images = os.path.getsize(file_path) // width // height
+    print(
+        f"{num_images} images of size {height}x{width} will be written to",
+        output_folder,
+    )
+
+    executor = ThreadPoolExecutor()
+    with open(file_path, "rb") as f:
+        for i in tqdm(range(num_images)):
+            img_data = f.read(width * height)
+            executor.submit(save_image, i, img_data, output_folder, width, height)
+            # save_image(i, img_data, output_folder, width, height)
+
+
+def init_parser():
+    parser = argparse.ArgumentParser()
+
+    parser.add_argument(
+        "--width", help="Width of a single frame", type=int, default=1920
+    )
+    parser.add_argument(
+        "--height", help="height of a single frame", type=int, default=1200
+    )
+    parser.add_argument("--source", help="Binary file.out", required=True)
+    parser.add_argument(
+        "--export",
+        help="Directotory to store raw single-channel .png files",
+        required=True,
+    )
+
+    return parser
+
+
+if __name__ == "__main__":
+    parser = init_parser()
+    args = parser.parse_args()
+
+    read_and_convert(args.source, args.export, args.width, args.height)

requirements.txt

@@ -1,3 +1,4 @@
 numpy==1.24.4
 opencv-python==4.8.0.76
 PyYAML==5.3.1
+tqdm