grapeot’s gists

grapeot / register.m

Last active December 23, 2015 07:18

MATLAB code to rectify the RGB and depth photos directly exported from Microsoft Kinect SDK. We assume the file `imgfn` stores the binary data directly exported from `ColorImageFrame.CopyPixelDataTo()`, `dfn` stores the binary data exported from `FusionFloatImageFrame.CopyPixelDataTo()`, and `idxfn` stores the binary data exported from the `Colo…

	function newImg = register(imgfn, dfn, idxfn)
	% rectify the color image according to the depth and the alignment
	% information from the data collection program
	% Consult http://grapeot.me/some-technical-details-about-kinectfusion.html for more details.

	%% read in the data
	fp = fopen(imgfn, 'r');
	imgRaw = fread(fp, 'char');
	imgRaw = reshape(imgRaw, 4, 640, 480);
	for i = 1: 3

grapeot / gitemail.php

Created September 24, 2013 22:22

A php script to send auto email notifications when a github repo is pushed. More details available at http://grapeot.me/automatic-email-notification-on-github-pushes.html.

	<?
	$r = 'This is an automatically generated message notifying you that someone pushed new commits to the github repo <project>.' . " Don't forget to do git pull before your editing.\r\n\r\nEdit summary: \r\n";
	$payload = $_REQUEST['payload'];
	preg_match_all('@"name":"([\w ]*?)","email@', $payload, $matches);
	$authors = $matches[1];
	preg_match_all('@"message":"([^"]*)"@', $payload, $matches);
	$messages = $matches[1];
	preg_match_all('@"url":"([^"]*)"@', $payload, $matches);
	$links= $matches[1];
	for ($i = 0; $i < count($messages) - 1; ++$i)

grapeot / install_numba.sh

Created November 19, 2013 23:37

install numba in Debian 7

	sudo apt-get install llvm cython python-pip python-llvm
	git clone https://github.com/numba/numba.git # Not sure whether this is necessary
	cd numba
	sudo pip install -r requirements.txt
	sudo pip install numba

grapeot / .zshrc

Last active August 29, 2015 13:55

my .zshrc

	# Path to your oh-my-zsh configuration.
	ZSH=$HOME/.oh-my-zsh

	# Set name of the theme to load.
	# Look in ~/.oh-my-zsh/themes/
	# Optionally, if you set this to "random", it'll load a random theme each
	# time that oh-my-zsh is loaded.
	ZSH_THEME="robbyrussell"

	# Example aliases

grapeot / .tmux.conf

Created February 28, 2014 18:09

	set -g default-terminal "screen-256color"
	set -g status-right '#[fg=black] #(hostname), %H:%M, #(uptime \| egrep -o "average.*"\|perl -pe "s\|average: \|\|")'
	setw -g automatic-rename

	# mouse mode
	set -g mode-mouse on
	setw -g mouse-select-window on
	setw -g mouse-select-pane on
	setw -g mode-keys vi

grapeot / extractFeatures.py

Created April 24, 2017 03:14

Use Caffe2 to extract features

	# Initial imports
	import os
	import sys
	import logging
	logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.INFO)

	from caffe2.proto import caffe2_pb2
	import numpy as np
	import skimage.io
	import skimage.transform

grapeot / track.py

Last active November 1, 2022 07:34

ZWO AM5自动跟踪红色滑块

	import zwoasi
	import cv2
	import numpy as np
	from simple_pid import PID
	from alpaca import Telescope

	def initializeCamera():
	zwoasi.init('./ASICamera2.dll')
	camera = zwoasi.Camera(0)
	controls = camera.get_controls()

grapeot / auto_focus.py

Created February 28, 2024 23:04

A reference implementation of auto-focusing by local entropy. https://yage.ai/auto-focus.html

	"""
	A class supporting the auto-focusing of a telescope.
	"""
	from astropy.io import fits
	from astropy.stats import sigma_clipped_stats
	from photutils.detection import DAOStarFinder
	from glob import glob
	from tqdm import tqdm
	from typing import List, Dict, Any
	from time import time

grapeot / gpt4o.py

Created June 17, 2024 15:11

使用GPT-4o识别云量

	import os
	import base64
	import requests
	import json

	def get_cloudiness(image_path):
	# 读取OpenAI API Key
	api_key = os.getenv('OPENAI_API_KEY')

	if not api_key:

grapeot / translate.py

Last active November 24, 2024 07:41

A reference implementation for performing offline inference using multiple GPUs, with each GPU hosting one instance of the model. Surprisingly, I couldn't find existing tools that easily support this. Therefore, I had to manually launch several server instances on different ports and use Ray's data set parallelization along with a server manager…

	# The VLLM servers were launched using something like:
	# CUDA_VISIBLE_DEVICES=2 vllm serve Qwen/Qwen2.5-32B-Instruct-GPTQ-Int4 --quantization gptq --max-model-len 4096 --port 8003

	from openai import OpenAI
	import ray
	from ray.util import ActorPool
	from tqdm import tqdm

	def read_file(file_path: str) -> list[str]:
	"""

Ya Ge grapeot