wei2912 commited on
Commit
25f0922
·
verified ·
1 Parent(s): f61db3d

feat(README): add usage instructions

Browse files
Files changed (1) hide show
  1. README.md +15 -3
README.md CHANGED
@@ -55,10 +55,17 @@ Project page: https://sphere-vlm.github.io/
55
  <img src="https://cdn-uploads.huggingface.co/production/uploads/66178f9809f891c11c213a68/g0___KuwnEJ37i6-W96Ru.png" alt="SPHERE dataset examples" width="400"/>
56
  </p>
57
 
58
- ## Dataset Preparation
59
 
60
- This version of the dataset was prepared by combining the [JSON annotations](https://github.com/zwenyu/SPHERE-VLM/tree/main/eval_datasets/coco_test2017_annotations) with the corresponding images from [MS COCO-2017](https://cocodataset.org).
61
- The script used can be found at `prepare_parquet.py`, to be executed in the root of [our GitHub repository](https://github.com/zwenyu/SPHERE-VLM).
 
 
 
 
 
 
 
62
 
63
  ## Dataset Structure
64
 
@@ -107,6 +114,11 @@ The data fields are as follows:
107
  - `source_img_id`: Source image ID in [MS COCO-2017](https://cocodataset.org).
108
  - `skill`: For reasoning tasks, a list of skills tested by the question, e.g. "count", "dist" (distance), "pos" (position), "shape", "size", "vis" (visual).
109
 
 
 
 
 
 
110
  ## Licensing Information
111
 
112
  Please note that the images are subject to the [Terms of Use of MS COCO-2017](https://cocodataset.org/#termsofuse):
 
55
  <img src="https://cdn-uploads.huggingface.co/production/uploads/66178f9809f891c11c213a68/g0___KuwnEJ37i6-W96Ru.png" alt="SPHERE dataset examples" width="400"/>
56
  </p>
57
 
58
+ ## Dataset Usage
59
 
60
+ To use this dataset, run the following:
61
+
62
+ ```python
63
+ from datasets import load_dataset
64
+
65
+ dataset = load_dataset("wei2912/SPHERE-VLM", "counting_only-paired-distance_and_counting")
66
+ ```
67
+
68
+ where the second argument to `load_dataset` is the subset of your choice (see [Dataset Structure](#dataset-structure)).
69
 
70
  ## Dataset Structure
71
 
 
114
  - `source_img_id`: Source image ID in [MS COCO-2017](https://cocodataset.org).
115
  - `skill`: For reasoning tasks, a list of skills tested by the question, e.g. "count", "dist" (distance), "pos" (position), "shape", "size", "vis" (visual).
116
 
117
+ ## Dataset Preparation
118
+
119
+ This version of the dataset was prepared by combining the [JSON annotations](https://github.com/zwenyu/SPHERE-VLM/tree/main/eval_datasets/coco_test2017_annotations) with the corresponding images from [MS COCO-2017](https://cocodataset.org).
120
+ The script used can be found at `prepare_parquet.py`, to be executed in the root of [our GitHub repository](https://github.com/zwenyu/SPHERE-VLM).
121
+
122
  ## Licensing Information
123
 
124
  Please note that the images are subject to the [Terms of Use of MS COCO-2017](https://cocodataset.org/#termsofuse):