chore: checkpoint"

Signed-off-by: ivelin <ivelin.eth@gmail.com>

ui_refexp.py

@@ -19,7 +19,7 @@ import csv
 import glob
 import os
 import tensorflow as tf
-
+import re
 import datasets
 
 import numpy as np
@@ -52,7 +52,8 @@ _LICENSE = "CC BY 4.0"
 # The HuggingFace dataset library don't host the datasets but only point to the original files
 # This can be an arbitrary nested dict/list of URLs (see below in `_split_generators` method)
 _DATA_URLs = {
-    "ui_refexp": "https://huggingface.co/datasets/ncoop57/rico_captions/resolve/main/captions_hierarchies_images_filtered.zip",
+    "ui_refexp": "https://storage.googleapis.com/crowdstf-rico-uiuc-4540/rico_dataset_v0.1/unique_uis.tar.gz"
+    # "https://huggingface.co/datasets/ncoop57/rico_captions/resolve/main/captions_hierarchies_images_filtered.zip",
 }
 
 _METADATA_URLS = {
@@ -64,8 +65,10 @@ _METADATA_URLS = {
 }
 
 
-def tfrecord2dict(raw_tfr_dataset: None):
-    """Filter and convert refexp tfrecord file to dict object."""
+def tfrecord2list(tfr_file: None):
+    """Filter and convert refexp tfrecord file to a list of dict object.
+    Each sample in the list is a dict with the following keys: (image_id, prompt, target_bounding_box)"""
+    test_raw_dataset = tf.data.TFRecordDataset([tfr_file])
     count = 0
     donut_refexp_dict = []
     for raw_record in raw_tfr_dataset:
@@ -76,9 +79,8 @@ def tfrecord2dict(raw_tfr_dataset: None):
         # print(f"feature keys: {example.features.feature.keys}")
         donut_refexp = {}
         image_id = example.features.feature['image/id'].bytes_list.value[0].decode()
-        image_path = zipurl_template.format(image_id=image_id)
-        donut_refexp["image_path"] = image_path
-        donut_refexp["question"] = example.features.feature["image/ref_exp/text"].bytes_list.value[0].decode()
+        donut_refexp["image_id"] = image_id
+        donut_refexp["prompt"] = example.features.feature["image/ref_exp/text"].bytes_list.value[0].decode()
         object_idx = example.features.feature["image/ref_exp/label"].int64_list.value[0]
         object_idx = int(object_idx)
         # print(f"object_idx: {object_idx}")
@@ -88,7 +90,7 @@ def tfrecord2dict(raw_tfr_dataset: None):
         object_bb["ymin"] = example.features.feature['image/object/bbox/ymin'].float_list.value[object_idx]
         object_bb["xmax"] = example.features.feature['image/object/bbox/xmax'].float_list.value[object_idx]
         object_bb["ymax"] = example.features.feature['image/object/bbox/ymax'].float_list.value[object_idx]
-        donut_refexp["answer"] = object_bb
+        donut_refexp["target_bounding_box"] = object_bb
         donut_refexp_dict.append(donut_refexp)
         if count != 3:
             continue
@@ -211,13 +213,15 @@ class UIRefExp(datasets.GeneratorBasedBuilder):
         # The `key` is here for legacy reason (tfds) and is not important in itself.
         # filter tfrecord and convert to json
 
-        with open(metadata_path, encoding="utf-8") as f:
-            files_to_keep = set(f.read().split("\n"))
+        metadata = tfrecord2list(metadata_file)
+        files_to_keep = set()
+        for sample in metadata:
+            files_to_keep.add(sample["image_id"])
         for file_path, file_obj in images:
-            if file_path.startswith(_IMAGES_DIR):
-                if file_path[len(_IMAGES_DIR): -len(".jpg")] in files_to_keep:
-                    label = file_path.split("/")[2]
-                    yield file_path, {
-                        "image": {"path": file_path, "bytes": file_obj.read()},
-                        "label": label,
-                    }
+            image_id = file_path.search("(\d+).jpg").group(1)
+            if image_id and image_id in files_to_keep:
+                label = file_path.split("/")[2]
+                yield file_path, {
+                    "image": {"path": file_path, "bytes": file_obj.read()},
+                    "label": label,
+                }