Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,74 @@
|
|
1 |
-
---
|
2 |
-
license: mit
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: mit
|
3 |
+
task_categories:
|
4 |
+
- video-classification
|
5 |
+
tags:
|
6 |
+
- baseball
|
7 |
+
- sports
|
8 |
+
- video-classification
|
9 |
+
- computer-vision
|
10 |
+
size_categories:
|
11 |
+
- 100K<n<1M
|
12 |
+
---
|
13 |
+
# Dataset Card for Picklebot130k
|
14 |
+
|
15 |
+
<!-- Provide a quick summary of the dataset. -->
|
16 |
+
|
17 |
+
130 thousand video clips of balls and strikes from MLB games from the 2016 season through the 2023 season.
|
18 |
+
![Example Clip](example.gif)
|
19 |
+
|
20 |
+
## Dataset Details
|
21 |
+
|
22 |
+
### Dataset Description
|
23 |
+
|
24 |
+
<!-- Provide a longer summary of what this dataset is. -->
|
25 |
+
The dataset consists of roughly 130 thousand video clips of balls and strikes in .mp4 format, resized to 224x224 resolution.
|
26 |
+
|
27 |
+
|
28 |
+
- **Curated by:** Henry Freed
|
29 |
+
- **License:** MIT
|
30 |
+
|
31 |
+
### Dataset Sources [optional]
|
32 |
+
|
33 |
+
<!-- Provide the basic links for the dataset. -->
|
34 |
+
|
35 |
+
- **Repository:** The original project that this dataset was compiled for can be found here on [github](https://github.com/hbfreed/Picklebot).
|
36 |
+
- **Demo:** The demo for a neural net trained on this dataset can be found here on [huggingface spaces](https://huggingface.co/spaces/hbfreed/picklebot_demo).
|
37 |
+
## Uses
|
38 |
+
|
39 |
+
<!-- Address questions around how the dataset is intended to be used. -->
|
40 |
+
The dataset was originally collected to call balls and strikes using neural networks. There are many other potential use cases, but they would almost certainly require relabeling. For more videos and more complete information about each pitch, see [Picklebot-2M](https://huggingface.co/datasets/hbfreed/Picklebot-2M).
|
41 |
+
|
42 |
+
## Dataset Structure
|
43 |
+
|
44 |
+
<!-- This section provides a description of the dataset fields, and additional information about the dataset structure such as criteria used to create the splits, relationships between data points, etc. -->
|
45 |
+
The dataset is structured as a .tar file of all the videos. The labels are contained in three .csv files (train, val, test). The .csvs are structured as follows:
|
46 |
+
|
47 |
+
date pitch mph spin_rate pitcher batter zone count inning pitch_result video_link filename video_length
|
48 |
+
2018-06-15 FF 94.8 2256 Gausman, Kevin(R) Shuck, JB(L) 6 2-2 Top 2 Called Strike https://baseballsavant.mlb.com/sporty-videos?playId=... clip_21926.mp4 200
|
49 |
+
|
50 |
+
Where the zone is the pitch location. The zones on Savant are presented as 1-9 for inside the zone, and 11-14 for outside the zone. We renumbered them by subtracting 1 from the inside zone values and 2 from the outside zone values, resulting in a continuous range of 0-12, where 0-8 represents inside the zone, and 9-12 represents outside the zone.
|
51 |
+
The dataset includes the following pitch labels: FF (Fastball/4-seam), SI (Sinker/2-Seam), FC (Cutter), CH (Changeup), FS (Split-finger), FO (Forkball), SC (Screwball), CU (Curveball), KC (Knuckle Curve), CS (Slow Curve), SL (Slider), ST (Sweeper), SV (Slurve), KN (Knuckleball), EP (Eephus), FA (Other), IN (Intentional Ball), and PO (Pitchout).
|
52 |
+
|
53 |
+
### Source Data
|
54 |
+
|
55 |
+
<!-- This section describes the source data (e.g. news text and headlines, social media posts, translated sentences, ...). -->
|
56 |
+
The source data were scraped from Baseball Savant's [Statcast Search](https://baseballsavant.mlb.com/statcast_search). It's a pretty powerful search page, and a lot of fun to play around with.
|
57 |
+
|
58 |
+
|
59 |
+
#### Data Collection and Processing
|
60 |
+
|
61 |
+
<!-- This section describes the data collection and processing process such as data selection criteria, filtering and normalization methods, tools and libraries used, etc. -->
|
62 |
+
|
63 |
+
After downloading the videos, they were cropped from 1280x720 to the middle 700x700 pixels and downsampled to 224x224 resolution at 15 fps (this can all be done using one ffmpeg command). Some of the longer clips where there was a lot of noise (shots of the crowd, instant replays, etc.) were trimmed (mostly by hand) down to a more manageable length.
|
64 |
+
|
65 |
+
#### Who are the source data producers?
|
66 |
+
|
67 |
+
<!-- This section describes the people or systems who originally created the data. It should also include self-reported demographic or identity information for the source data creators if this information is available. -->
|
68 |
+
[Baseball Savant](https://baseballsavant.mlb.com/) and MLB/the broadcasters (whoever it is) originally created the videos.
|
69 |
+
|
70 |
+
|
71 |
+
## Bias, Risks, and Limitations
|
72 |
+
|
73 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
74 |
+
It's important to note that only balls and called strikes were collected. No swinging strikes, foul balls, hit by pitches, or anything else are included in the dataset. Additionally, most pitchers and batters are right handed, and nothing was done to try and balance that in this dataset.
|