On this video, I discuss the start of planning a workflow of how I think the models will collect data, train, and test models.
Welcome to my machine learning blog on the next series of videos I will be discussing how I plan to use machine learning framework with other technology frameworks to create applications please join me by subscribing to my blog and ask questions or give answers to the problems I encounter on the concepts I am built all right operations investigations in the use of features with sickness and this will our dodo are data sets it go be broken down sorry a data source sores is are coming from investigators either from investing years in video For audio so that raise investigations and we party look at that later on is you have you guys in the paranormal investigating paranormal of trying to check ghosts zone more hearing disembodied voices summit taking pictures and they see formations operations of permissions in investigating the source of the sources of try to identify so what happened is that some his videos are always hours on the QB from cameras it takes time to process this video and maybe the letter, show you what I mean in history tablets so Michael for operations investigations a data set that comprises the video process audio for these individuals so they can easily hold related to tax some of these analogies for them so they can focus and more details in case the tool does not detected right so the data server comfort investigators it into their own cameras all when there don’t good with the caption that Of the medial and did so without looking at is my will cameras and investigations, that’s who get data source against the video and audio video video and audio mother searches YouTube videos on for a from my channels that school investigations for explorations so if you haven’t gone through the webpage or website built on the fundamental operations investigations of these are standing together from year this is a brother their two brothers that have a channel and are actually there in or near urban ghosts and old urban explorations so the going to an urban environment this CD in my going to Cemetery and exposed Cemetery day and night on the Michael to an abandoned building summit that and this VUR this video capturing, things courses is just a format where capture the video and the biggest stills and is just a matter of you just showing you the environment and somebody and am looking at is to identify the environment that therein right now again that be can be go of the the outside a be insight and for Is here looks at this particular radio is an Cemetery song want to build to detect tombstones and these tombstones were equal to a Cemetery so evident in a coastal point that I need to separate Madia says between an outside environment and an internal environment I can do that and then of course that would later and I could identify objects for example in a Cemetery is identify headstones started the identifying unit of the years in the headstones are like that are not yet but the biggest thing is to identify it will have bugs there from the outside were bugs are and insight is falling because of an operation can be one or and you can disqualify a norm as either being a bug or dust particle or something that some apartment isn’t am doing is a process the video to get tax and I have described tax year and the transcribed taxes from the video source of the taxes on doing speech to text recognition of some of these video sources environment year and in some D’Souza have a lot of videos some then I was one of the majors delays for me for having this sorrel how to them doing this is all prices and his data so I have several computers and number to draw this configuration that have right now so I several computers doing the speech to text recognition audio processing so goal back here is where the features on half is to isolate images where leave will be using for classification and audio detection so if you haven’t already gone to the website them to happening here yes I have as a posted this article here where it goes into IC audio using CNN collision of your networks and references this on this article in that you can use us a spectrogram on in the speck from spectrogram format at abilities it into training accomplished network machine learning is a spectrogram and you so and the course spectrogram has you can think of she the time and it’s an actual image again using an instrument to train our machine photos so we go back to my original article here in Canada explains this life I audio portion looks like by saying testing 123 testing 123 said I create these and and and creating a data set so what could happen at the end is that this videos out Free from these guys talking something that would mentioned that when they’re on the cemeteries that here makes soul one who is as part of my transcribed text is a transcribed his videos on my prescriptions are not accurate because I’m itself constrained as well salt Harry W some of the audio this video sometimes but it’s really captures certain keywords the what he’s these future numbers 40s hundred and are saying I heard this for each year that or that you say something that will be let keywords within the video file to be ever captured audio and audio but some video cards for example abilities and’s or maybe they see something from their eyes. Shadow people such people are discussed find shadow people will get classification that he to label to those will be some of the pictures that want to target Sally’s private text for this different project so to be a isolate images for confiscations and then also autodetection so previously for him production environment you will him the user to upload a video and application processes both video and audio detection of anomalies mildly set in well in achieving this group is just think it will release and immediate Douglas was the training boundary is for them to set a have not work the video and later on show you what I’m the I might not have is a with applicable cameras and even on my have you said when I go to investigation the people not all that will be able to process the amendment have time to process the video help me identify on August so when I’m hoping to have is that they getting have co-cameras may be running in one investigation I wanted to be a virtues on these cameras the get the data to get a different good was videophiles uploaded to the system in the system actually go through the video the audio and text book section heart attack but in my to imitate the money quicker than sometimes will identify resource section blessed second investigations in our but it’s just me that takes long time to process this data in With identify it understand with So at the End That They Victory Machinery to Identify and Him Production Environment Is to Have the Will with the Casket up with Video and Audio Detection the Benefits More identify It’s an Insect Dust or It’s an Actual Operation or Google Connect so That Stengel Make Is That That This This Will Be for the Space to Buildings so Sometimes Some of These Guys and How Can Imagine to Discuss That Consumes A Lot Of Their Time and
My Dearest Artist of the Because It’s Really Time Consuming Is Not Only Do They Have To These Captures That the O’clock I. But It Some of the Futuristic and Visiting with You to Visit of the Mentioned Here Are It’s on My Website As of the Schedule and and You Have Baked Pork Chops so the Discontinuance of the the You the Users or Other Subscribers to Help Identify Certain Things of His Abilities Are Easy to Identify Because They Go out There Experience Said and the United States the Tagging or Even Mention It into in the Video to Get That Her Dad of the Just Heard Is a Heard Not so They Do a Quick Edit and Publish and You to in Hopes That Their Subscriber Base Wife to Identify from It but If You to Go to Any Other Any of Their Vehicles and Comments As Part of a Huge and Smith This Information out Is Confidence out and Get at Times so of the Longest to Have That Process Myself Identify and to Apply for the Machine Is Actually Going to Be a Finding of Anomalies That This Is Doing the Job so Anyways the Post Here on That He POSTNET It’s Been Assigned This Health Facilities at Several Search Analytics Unit Commons Area Is the Categories Have Broken down That You Have Some Explorers after Investigators Actually Investigating This Is Our Experiences His Videos That I Had Created Based on My Experiences and Then Have Further Broken down to These Unknowns for the Explored This Increase Has Been Initiated to Us Are Likely Type Beautiful Vessels Topics but That Is Again You Get Started Back in February and so the Shortly Trying Get the Status Said so to Increase My Data Set and Adhere, You in Here on Your Bids Ghost Hunters Paranormal Investigators in Society Is/Directory Listings so the Corporate When Crawler That Will Index so I Do Have Will WebCrawler Engine of These Show You Guys That As Well As As Part of Getting the Data Source for This What Does Weber Color Do This I Go to Websites I Enjoyed Some of These Societies of Second Facebook Connect Have a Website There’s There’s Directory Listings and Again Those Websites and Have the WebCrawler and Initially Start Some and Then across a More and Then Again Additional Web Domains Associated with That Website and Show You How I Do That As Part of a Separate and Independent Be Part of Identifying Our Data Sources so Having Said That What I Do What You Know These Websites and Web Probably His Images Identifying Video and Audio Files As Well so Hoping That That Would Generate Another Data Source for Me Especially since the YouTube Videos That Has To Be with Explorations Right Now They Are Some You to Channels That He Do Have Some the but That Will Be Another Source Began in May Features Will Will Have Some Inputs Some Outlets so This Is Very Common Here I Guess I Was a User Here with That User/Subscribers And Entire That Insipid User/Subscriber and Here I Guess It in General Be Data Source Software (Butterworth Repository Distributed Database to Sell the Data Source/Repository so the Star to Identify Where Data Sources That Were Expected, so One at Step You Invest That Is so First Accident of Our Data Source Based on Our Features of Salt Harry Identifying Number Going to Be Chosen in Easy to Individual Steps but First I Would Only Do Is Going to Move This over Been Next so They on Asked and Development It’s All Consider Myself As a User As Well Subscriber Is Gather and so Were the Processes Identify Data Sources Should Be Very Day Gather the Data so Began to Be Users/They Considered a WebCrawler Join Together and It Is Certainly Start Labeling Testifying Saw the Labeling Testifying to Build a Web Interface to Be Able to Do Again Break into the Steps so One of the Things You Don’t Have Been Forgot to Mention You May Have Mentioned the Step Was on Increases and for a Decision Point and the Property Jealous They Only Achieve At Least 80% Confidence 80% Confidence so Him It’s Not Then Of Course It’s Google Back Here to Collect Additional Data and Then It Is Then It Is Them to Prepare the Data Sets for Deployments so That Will Be a Decision Point Me so Hoping There’s Any Other Position Points You Them Might Be Individually but to Noon Not Have Never You but This Is a High-Level View We Want At Least for Me I Want to Be Able to Have the Machine to Be Able to Be 80% Confidence That an Object Exists a Bit Them in the Gold Could Be Coming Were Not in Many Mission-Critical Times Five Separate Your Life Saving Time Project Saw It Again I Get 80% Confidence the Depictions of the Super Happy and It’s Not an “Push in Fact That and Additional Data Sources to Improve the Confidence Level the 80% Confidence Objection to His Making Sure That Additional Data Corruption so For Example in Detecting an Order for Can Be Broken down to Make the above or Dust Were an Anomaly to You Have Three and It’s an Actual Blogging Essays It’s 80% of Its Dust Then That Is Data Corruption so We Want to Have That Decision Point Right Here That We to Go Back to Our Data Be Together Additional Pay Sources and or Your Check Our Labeling and Classification Make Sure That Additional Corruption Injected into Our Data Sense so We Definitely Best That Been There so I’d Highly Reveal All This Is This This Is How and a Half for Now We Might Change Companies to It All of the Defaces Is Long and I Think We’re Good Here for Today It to Create More so for This Is for the Save Its into My Got Two Different Ones Year the Human News Channels Putting Here and Save Its into My Repository for Another Severity Women Image Machine Learning Workflow for Floor Things Soon Colonists Will Consist of Several Pages to Change the on at the Things He Thought/Should He His Face He Was Not a Single Development That We Get to This Training Here I Haven’t Decided yet That the Other Party to Use My CPUs to Start but Is This Girl Does Grow I Might Use Google Cloud Services to the Actual Processing to Speak to the Trains Are Guys Been Wanting Continue Bit of December Here All Close to This Video Blog Are My Personal Opinion and They Don’t in Any Way Reflect Opinions of My Poor All Materials Posts and Advice from This Site Are Informational Researching and for Testing Purposes of the to Use Them at Your on Responsibilities I Not in Any Way Responsible for Any Damage Done by Following Posts Advice Tutorials and Articles from This Blog My Machine Learning Blog Man Not Own Some of the Continent Percent