| import onnx | |
| onnx_model = onnx.load('./input.onnx') | |
| #Rename 'inp' to 'inst' | |
| endpoint_names = ['inp', 'inst'] | |
| for i in range(len(onnx_model.graph.node)): | |
| for j in range(len(onnx_model.graph.node[i].input)): | |
| if onnx_model.graph.node[i].input[j] == endpoint_names[0]: |
| from transformers import AutoTokenizer | |
| from transformers import LlamaTokenizerFast | |
| #tokenizer_yam = AutoTokenizer.from_pretrained("yam-peleg/Hebrew-Gemma-11B-V2") | |
| tokenizer_grok = LlamaTokenizerFast.from_pretrained('Xenova/grok-1-tokenizer') | |
| tokenizer_gemma = AutoTokenizer.from_pretrained("google/gemma-7b-it") | |
| tokenizer_aya101 = AutoTokenizer.from_pretrained("CohereForAI/aya-101") | |
| tokenizer_gpt2 = AutoTokenizer.from_pretrained("gpt2") |
| import torch | |
| import gradio as gr | |
| from threading import Thread | |
| from transformers import AutoModelForSeq2SeqLM, AutoTokenizer, TextIteratorStreamer | |
| checkpoint = "CohereForAI/aya-101" | |
| tokenizer = AutoTokenizer.from_pretrained(checkpoint) | |
| model = AutoModelForSeq2SeqLM.from_pretrained(checkpoint, device_map='auto', torch_dtype=torch.bfloat16) |
| # crapify_images.py | |
| # Go over a folder of images and reduce the quality and/or add noise to some of them (Useful for making a Pix2Pix model more resilient) | |
| # @Norod78 | |
| import skimage | |
| import skimage.io | |
| import skimage.io._plugins.pil_plugin as pp | |
| import numpy as np | |
| from PIL import Image |
| To enhance readability, I'll add more line breaks within the sections to make the text easier to follow. Here's a revised version with additional spacing: | |
| "You are ChatGPT, a large language model trained by OpenAI, based on the GPT-4 architecture." | |
| "Image input capabilities: Enabled" | |
| "Conversation start date: 2023-12-19T01:17:10.597024" | |
| "Deprecated knowledge cutoff: 2023-04-01" |
This Python script utilizes the dlib library for face detection and facial landmarks to detect and crop a human head in an image. The detect_and_crop_head function takes an input image file, detects the face using a pre-trained 5-point facial landmarks model, and crops the image to include the entire head with a 1:1 aspect ratio. The cropping is adjusted based on the roll and yaw angles calculated from the facial landmarks.
This Python script leverages the OpenCV library for face detection and the PIL (Pillow) library for image processing to detect and crop a human head in an image. The detect_and_crop_head function takes an input image file, detects the face using a pre-trained Haar Cascade classifier, and crops the image to include the entire head with a 1:1 aspect ratio. The cropping is achieved by specifying a factor that determines the size of the cropped region around the detected face. Adjust the factor parameter based on the specific requirements of your images.
input_image_path = "path/to/your/input/image.jpg"
output_image_path = "path/to/save/cropped/head.jpg"
detect_and_crop_head(input_image_path, output_image_path, factor=1.5)