proof_of_concept.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\pithy\\anaconda3\\lib\\site-packages\\pandas\\core\\computation\\expressions.py:20: UserWarning: Pandas requires version '2.7.3' or newer of 'numexpr' (version '2.7.1' currently installed).\n",
      "  from pandas.core.computation.check import NUMEXPR_INSTALLED\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "03332f84d40f4a5ea3cc8a8ef0a27e78",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='config.json'), FloatProgress(value=0.0, max=4165.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\pithy\\anaconda3\\lib\\site-packages\\huggingface_hub\\file_download.py:149: UserWarning: `huggingface_hub` cache-system uses symlinks by default to efficiently store duplicated files but your machine does not support them in C:\\Users\\pithy\\.cache\\huggingface\\hub\\models--microsoft--trocr-base-handwritten. Caching files will still work but in a degraded version that might require more space on your disk. This warning can be disabled by setting the `HF_HUB_DISABLE_SYMLINKS_WARNING` environment variable. For more details, see https://huggingface.co/docs/huggingface_hub/how-to-cache#limitations.\n",
      "To support symlinks on Windows, you either need to activate Developer Mode or to run Python as an administrator. In order to see activate developer mode, see this article: https://docs.microsoft.com/en-us/windows/apps/get-started/enable-your-device-for-development\n",
      "  warnings.warn(message)\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "19b9fc988bfa43e0bb88e4360384af48",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='model.safetensors'), FloatProgress(value=0.0, max=1333384464.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Some weights of VisionEncoderDecoderModel were not initialized from the model checkpoint at microsoft/trocr-base-handwritten and are newly initialized: ['encoder.pooler.dense.bias', 'encoder.pooler.dense.weight']\n",
      "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "be89cba393e54f09af2135f59930da63",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='generation_config.json'), FloatProgress(value=0.0, max=190.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "534e24f4cc9148bcb0612bccdd865b5a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='tokenizer_config.json'), FloatProgress(value=0.0, max=1118.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1601a0ccfaf1460d8b11332e8ed13b65",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='vocab.json'), FloatProgress(value=0.0, max=898822.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "bf2a6a0c996d43c0ac0a274c6c42fba3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='merges.txt'), FloatProgress(value=0.0, max=456318.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "38bc148b24654cd7992fa0fd5e64996d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='special_tokens_map.json'), FloatProgress(value=0.0, max=772.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b815edf949ec4e199a3eefc666e964b9",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='preprocessor_config.json'), FloatProgress(value=0.0, max=228.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Could not find image processor class in the image processor config or the model config. Loading based on pattern matching with the model's feature extractor configuration.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\pithy\\anaconda3\\lib\\site-packages\\transformers\\models\\vit\\feature_extraction_vit.py:28: FutureWarning: The class ViTFeatureExtractor is deprecated and will be removed in version 5 of Transformers. Please use ViTImageProcessor instead.\n",
      "  warnings.warn(\n"
     ]
    }
   ],
   "source": [
    "# Use a pipeline as a high-level helper\n",
    "from transformers import pipeline\n",
    "\n",
    "pipe = pipeline(\"image-to-text\", model=\"microsoft/trocr-base-handwritten\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformers import TrOCRProcessor, VisionEncoderDecoderModel\n",
    "from PIL import Image\n",
    "import requests\n",
    "import os\n",
    "import datetime"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7cc594c249a6463bab2bb00400a33856",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='preprocessor_config.json'), FloatProgress(value=0.0, max=228.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\pithy\\anaconda3\\lib\\site-packages\\huggingface_hub\\file_download.py:149: UserWarning: `huggingface_hub` cache-system uses symlinks by default to efficiently store duplicated files but your machine does not support them in C:\\Users\\pithy\\.cache\\huggingface\\hub\\models--microsoft--trocr-base-stage1. Caching files will still work but in a degraded version that might require more space on your disk. This warning can be disabled by setting the `HF_HUB_DISABLE_SYMLINKS_WARNING` environment variable. For more details, see https://huggingface.co/docs/huggingface_hub/how-to-cache#limitations.\n",
      "To support symlinks on Windows, you either need to activate Developer Mode or to run Python as an administrator. In order to see activate developer mode, see this article: https://docs.microsoft.com/en-us/windows/apps/get-started/enable-your-device-for-development\n",
      "  warnings.warn(message)\n",
      "Could not find image processor class in the image processor config or the model config. Loading based on pattern matching with the model's feature extractor configuration.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "70c4e99eeadf4d2da9de4c880b5fed34",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='tokenizer_config.json'), FloatProgress(value=0.0, max=1118.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a8cea6d39f4e4d72bd06b45757e80ca0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='vocab.json'), FloatProgress(value=0.0, max=898822.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d7c6e41fc0974d16858f18d715980a28",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='merges.txt'), FloatProgress(value=0.0, max=456318.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c370fe3060ae4f84ab127f81077098a8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='special_tokens_map.json'), FloatProgress(value=0.0, max=772.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b7e2faf8674442a491394c50f1b4328e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='config.json'), FloatProgress(value=0.0, max=4207.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "46714854d37d424a93a8848323076f41",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='pytorch_model.bin'), FloatProgress(value=0.0, max=1537279879.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Some weights of VisionEncoderDecoderModel were not initialized from the model checkpoint at microsoft/trocr-base-stage1 and are newly initialized: ['encoder.pooler.dense.weight', 'encoder.pooler.dense.bias']\n",
      "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "47a13fd0aeb64e97939747d84d488d93",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='generation_config.json'), FloatProgress(value=0.0, max=190.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\pithy\\anaconda3\\lib\\site-packages\\transformers\\generation\\utils.py:1355: UserWarning: Using the model-agnostic default `max_length` (=20) to control the generation length. We recommend setting `max_new_tokens` to control the maximum length of the generation.\n",
      "  warnings.warn(\n",
      "Could not find image processor class in the image processor config or the model config. Loading based on pattern matching with the model's feature extractor configuration.\n",
      "Some weights of VisionEncoderDecoderModel were not initialized from the model checkpoint at microsoft/trocr-base-stage1 and are newly initialized: ['encoder.pooler.dense.weight', 'encoder.pooler.dense.bias']\n",
      "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n",
      "Could not find image processor class in the image processor config or the model config. Loading based on pattern matching with the model's feature extractor configuration.\n",
      "Some weights of VisionEncoderDecoderModel were not initialized from the model checkpoint at microsoft/trocr-base-stage1 and are newly initialized: ['encoder.pooler.dense.weight', 'encoder.pooler.dense.bias']\n",
      "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n",
      "Could not find image processor class in the image processor config or the model config. Loading based on pattern matching with the model's feature extractor configuration.\n",
      "Some weights of VisionEncoderDecoderModel were not initialized from the model checkpoint at microsoft/trocr-base-stage1 and are newly initialized: ['encoder.pooler.dense.weight', 'encoder.pooler.dense.bias']\n",
      "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n",
      "Could not find image processor class in the image processor config or the model config. Loading based on pattern matching with the model's feature extractor configuration.\n",
      "Some weights of VisionEncoderDecoderModel were not initialized from the model checkpoint at microsoft/trocr-base-stage1 and are newly initialized: ['encoder.pooler.dense.weight', 'encoder.pooler.dense.bias']\n",
      "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n",
      "Could not find image processor class in the image processor config or the model config. Loading based on pattern matching with the model's feature extractor configuration.\n",
      "Some weights of VisionEncoderDecoderModel were not initialized from the model checkpoint at microsoft/trocr-base-stage1 and are newly initialized: ['encoder.pooler.dense.weight', 'encoder.pooler.dense.bias']\n",
      "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n",
      "Could not find image processor class in the image processor config or the model config. Loading based on pattern matching with the model's feature extractor configuration.\n",
      "Some weights of VisionEncoderDecoderModel were not initialized from the model checkpoint at microsoft/trocr-base-stage1 and are newly initialized: ['encoder.pooler.dense.weight', 'encoder.pooler.dense.bias']\n",
      "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n",
      "Could not find image processor class in the image processor config or the model config. Loading based on pattern matching with the model's feature extractor configuration.\n",
      "Some weights of VisionEncoderDecoderModel were not initialized from the model checkpoint at microsoft/trocr-base-stage1 and are newly initialized: ['encoder.pooler.dense.weight', 'encoder.pooler.dense.bias']\n",
      "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n",
      "Could not find image processor class in the image processor config or the model config. Loading based on pattern matching with the model's feature extractor configuration.\n",
      "Some weights of VisionEncoderDecoderModel were not initialized from the model checkpoint at microsoft/trocr-base-stage1 and are newly initialized: ['encoder.pooler.dense.weight', 'encoder.pooler.dense.bias']\n",
      "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n"
     ]
    }
   ],
   "source": [
    "then = datetime.datetime.now()\n",
    "outputs = []\n",
    "for filename in os.listdir(\"images\"):\n",
    "    image = Image.open(f\"images/{filename.lower()}\").convert(\"RGB\")\n",
    "\n",
    "    processor = TrOCRProcessor.from_pretrained('microsoft/trocr-base-stage1')\n",
    "    model = VisionEncoderDecoderModel.from_pretrained('microsoft/trocr-base-stage1')\n",
    "    pixel_values = processor(images=image, return_tensors=\"pt\").pixel_values\n",
    "\n",
    "    generated_ids = model.generate(pixel_values)\n",
    "    generated_text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]\n",
    "    outputs.append(generated_text)\n",
    "now = datetime.datetime.now()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "11:52:41.587853 11:54:51.945369\n",
      "['\" Sang! When will it be, my turn again! I\\'m', 'very lonely why did they outcome Patrick Page', 'so much? Francesca said she has the \" I could', 'Do that \" gone - I have the related #', '\" I Need to make this my life\\'s work \" gene. \" Is', 'that just what being inspired is? Are I inspired', 'to write a musical now? IDK. Might go to a', 'baseball game later, but might be too sad going', \"by myself, so maybe. I'll stay in and rematch\"]\n"
     ]
    }
   ],
   "source": [
    "# trocr-base-handwritten\n",
    "print(then.time(), now.time())\n",
    "print(outputs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "15:10:12.422150 15:13:26.581881\n",
      "[\"song! When will it be my turn again! I'm\", 'very lonely why did they autotime Patrick Page', 'so much? Francesca said she has the \"I could', 'Do That \" gone - I have the related se', '\"I Need to Make This My Lites Work\" gene. Is', 'that just what being inspired is? Am I impred', 'to write a mvical now? IDK. Might go to a', 'baseball game later, but might be too sad going', \"by myself, so maybe I'll stay in and rewatch\"]\n"
     ]
    }
   ],
   "source": [
    "# trocr-base-stage1\n",
    "print(then.time(), now.time())\n",
    "print(outputs)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}