xvdp

"""
Conversion from WGS-84 to Transverse Mercator
WGS-84 is World Geodetic System data from 1984, utilized by GPS standard.
Transverse mercator is a general Mercator projection (projection of the oblate spheroid of earth onto a cylinder) that is used in most mapping systems. 
These short codes allow user to specify center latitude and longitude.

Contains a simpler projection assuming earth is a sphere
>>> tmercator_meters_to_gps(coords, center)
>>> tmercator_gps_to_meters(coords, center)

xvdp

#!/bin/bash
# change cache paths for huggingface, torch, openai (uses ~/.cache/ by default)
# use: set_model_caches.sh /mnt/my/model/cache/path

cache_path=$1
# Check if the argument is provided
if [ -z "$cache_path" ]; then
  echo "Error: Cache path argument is missing."
  echo "Usage: $0 /path_to_cache"
  exit 1

xvdp

""" @xvdp
Simple solution  to Show frame number and frame time with VLC or other video players
Make a subtitle file

caveats:
  only tested in linux
  requires ffmpeg / only tested with ffmpeg 4.3
  only tested with limited number of .mov and .mkv files
  it may fail if ffprobe reports nb_frames, DURATION or frame_rate with keys not parsed here.
  

xvdp

"""@xvdp
reciprocals for torch.cumsum and torch.cumprod
I noticed that torch has cumsum and cumprod but not their reciprocals
even thought cumdif and cumdiv have meanings in analysis and probability
and are routinely used.

Are these interpretations correct?

> cumsum can be thought of as a discrete integral
> cumdif as discrete derivative

xvdp

xvdp

xvdp

"""
>>> import interpolation_diff as idf
>>> algs = ("pil", "tensorflow") # ("pil", "torch")
>>> save = "/home/z/share/benchestchirescat_%s_%s"%algs
>>> idf.test_dif(dtype="float32", mode="bilinear", show=True, algs=algs, crop=(170,370, 180, 380), save=save)

# re https://discuss.pytorch.org/t/nn-interpolate-function-unexpected-behaviour/106684/6
"""
import io
import os

xvdp

xvdp

"""
Perceptual Loss Toy Implemntation from https://arxiv.org/pdf/1603.08155.pdf
"""
import os.path as osp
import torch
from torch import Tensor
import torch.nn as nn
import torch.nn.functional as F
from torchvision import models

xvdp

""" I was writing a dataloader from a video stream. I ran some numbers.
# in a nutshell. 
-> np.transpose() or torch.permute() is faster as uint8, no difference between torch and numpy
-> np.uint8/number results in np.float64, never do it, if anything cast as np.float32
-> convert to pytorch before converting uint8 to float32
-> contiguous() is is faster in torch than numpy
-> contiguous() is faster for torch.float32 than for torch.uint8
-> convert to CUDA in the numpy to pytorch conversion, if you can.
-> in CPU tensor/my_float is > 130% more costly than tensor.div_(myfloat), however tensor.div_()
does not keep track of gradients, so be careful using it.