Context:
Like many others, I had been putting off digitizing, sorting and categorizing my photo library for years! My grandmother past away midway through 2021, among other things, I was asked to find as many photos of my grandmother as I could.
Grief can do weird things to people, in my case it motivated me to sort my unsorted hoard into a scalable, manageable hoard.
Step 0: Assess the situation:
Ever since I was a little kid, my mother loved to take photos. She has tons of them around the house. Aside from the yet to be digitized physical photos, (Future project, maybe when my next grandparent passes~) the vast majority of my digital family photos existed as direct memory card "copy and pastes" into folders named"Unsorted" number one through a hundred. There was no attempt at a defined structure and was a mess of duplicates.
At first I was worried, then I realized "I get to decide the structure, I have to make a plan."
Step 1: Plan your structure:
I did some research and found that this is a hotly debated subject. Based on subject, camera, person in the photo, resolution, event, etc.
My point is, you need to make a decision. I personally decided on sorting into folders based on the year. This was an easy choice for me as the hoard I was sorting had EXIF metadata intact! In the next few steps I will be using this metadata to do the initial heavy lifting.
Step 2: Bulk rename, with automation!
As my hoard had been collected using various digital cameras and cell phones over the years, it was a mess of naming standards. I needed to rename every single file. I was not going to do this by hand, and neither should you!
I decided that as all of my photos had the date they had been taken embedded in metadata, I needed to find a way to bulk rename based on the existing data.
Introducing Irfanview. This program is a Swiss army knife of a photo tool. In this context, I am specifically after their Batch Renaming tool. By default this tool is set to make a copy of your existing files, rename them based on a pattern and move them to a dump file after the fact. I highly recommend you test this multiple times before you decide on the naming pattern you want!
I decided that I wanted to format my photo names as: *Photo - Year - Month - Day (Photo#)*I was able to accomplish this using Name pattern: *Photo - $E306(%Y-%m-%d)*They have pretty good documentation on setting up naming patterns. Not to recommend just "RTFM" but its basically just that, find the value you want, declare it and fire.
After doing a bulk rename, you can simple sort based on name and move your photos into their respective year.
It can also de-duplicate your photos, depending on your hoard, this can be very useful.
Note on Irfanview vs Digikam: I am aware that DigiKam can metadata rename and de-duplcate as well. I was unaware of this when I did my initial rename. Personally I prefer Irfanview as its more modular for renaming.
Step 4: Choose your weapon!
Both fortunately and unfortunately, there are a large number of expensive and proprietary software for helping with family photo management. This made it quite difficult to find a tool set that worked for my use case.
I don't want to take away from the work these programs do, however Cloud based, proprietary and with a recurring fee is not the Data Hoarder way. In my testing each one was either missing a feature or relied on a data format you would be locked into. I wanted a piece of software that would decrease my management overhead without breaking the bank or harvesting a pound of flesh each month, either in money or by collecting my data to do it.
Introducing DigiKam. This program is an opensource beast of a photo tool. Built to handle large libraries with a simple (I didn't say modern) interface and the ability to ingest any photo you give it, makes it my weapon of choice.
I do however want to give a shout out to PhotoPrism, if they had face detection and recognition released already, I probably would have settles with them first.
Step 5: Teaching Sand to see
Out of the box Digikam takes a little bit of configuration. I would suggest sticking with the default SQLite databases in most cases. Just be aware that if you are dealing with an especially large Hoard, this will make face recognition slower as the database gets larger. One of the startup questions is regarding Metadata write back, I highly suggest you enable this feature as this will allow you to more easily move out of Digikam and into a different platform more easily. (Like PhotoPrism) That being said, test with a sample first.
Once Digikam is configured, go to the "People" tab. Digikam will need to run an initial scan to find all of the faces in your Hoard, this can and will take time. I have found that the default sensitivity is pretty good to start, but I recommend enabling "YOLO v3 Detection" as it gives better results. "Work on All processor cores" seems to be broken at time of writing.
Let the facial detection do its thing, future scans are significantly faster as previous results are merged together resulting in a lower workload the more it is run. If you feel the need, you can force a full scan again.
Step 6: Finding Grandma
Once Digikam finishes finding faces, you will need to manually tell it who some of the faces are. The more photos you tag initially, the larger the starting data set will be and the higher accuracy the Individual Person recognition will be. (I had a lamp that it insisted was my grandfather at initial stages. Really odd but made me laugh way harder than I should have.)
After the first run, you will need to confirm the photos it thinks is the person you are working on. This step will be repeated over and over until a large enough sample is collected that you will eventually just be bulk confirming photos.
I encourage you to play with the sensitivity values while doing this as it can help very quickly. Running bulk face detection, people detection and confirming is the vast majority of the labor needed in this process.
Step 7: Profit
After confirming the photos to a level you are happy with, you are left with a date sorted, person tagged hoard of photos which can easily be searched in normal file viewers and is modular enough to be moved between system without having to worry about software dependencies, proprietary formats or data loss from a company you trusted to keep your hoard safe.
Best of all, the management overhead on this is pretty low and gets easier every time you do it.
Like many others, I had been putting off digitizing, sorting and categorizing my photo library for years! My grandmother past away midway through 2021, among other things, I was asked to find as many photos of my grandmother as I could.
Grief can do weird things to people, in my case it motivated me to sort my unsorted hoard into a scalable, manageable hoard.
Step 0: Assess the situation:
Ever since I was a little kid, my mother loved to take photos. She has tons of them around the house. Aside from the yet to be digitized physical photos, (Future project, maybe when my next grandparent passes~) the vast majority of my digital family photos existed as direct memory card "copy and pastes" into folders named"Unsorted" number one through a hundred. There was no attempt at a defined structure and was a mess of duplicates.
At first I was worried, then I realized "I get to decide the structure, I have to make a plan."
Step 1: Plan your structure:
I did some research and found that this is a hotly debated subject. Based on subject, camera, person in the photo, resolution, event, etc.
My point is, you need to make a decision. I personally decided on sorting into folders based on the year. This was an easy choice for me as the hoard I was sorting had EXIF metadata intact! In the next few steps I will be using this metadata to do the initial heavy lifting.
Step 2: Bulk rename, with automation!
As my hoard had been collected using various digital cameras and cell phones over the years, it was a mess of naming standards. I needed to rename every single file. I was not going to do this by hand, and neither should you!
I decided that as all of my photos had the date they had been taken embedded in metadata, I needed to find a way to bulk rename based on the existing data.
Introducing Irfanview. This program is a Swiss army knife of a photo tool. In this context, I am specifically after their Batch Renaming tool. By default this tool is set to make a copy of your existing files, rename them based on a pattern and move them to a dump file after the fact. I highly recommend you test this multiple times before you decide on the naming pattern you want!
I decided that I wanted to format my photo names as: *Photo - Year - Month - Day (Photo#)*I was able to accomplish this using Name pattern: *Photo - $E306(%Y-%m-%d)*They have pretty good documentation on setting up naming patterns. Not to recommend just "RTFM" but its basically just that, find the value you want, declare it and fire.
After doing a bulk rename, you can simple sort based on name and move your photos into their respective year.
It can also de-duplicate your photos, depending on your hoard, this can be very useful.
Note on Irfanview vs Digikam: I am aware that DigiKam can metadata rename and de-duplcate as well. I was unaware of this when I did my initial rename. Personally I prefer Irfanview as its more modular for renaming.
Step 4: Choose your weapon!
Both fortunately and unfortunately, there are a large number of expensive and proprietary software for helping with family photo management. This made it quite difficult to find a tool set that worked for my use case.
I don't want to take away from the work these programs do, however Cloud based, proprietary and with a recurring fee is not the Data Hoarder way. In my testing each one was either missing a feature or relied on a data format you would be locked into. I wanted a piece of software that would decrease my management overhead without breaking the bank or harvesting a pound of flesh each month, either in money or by collecting my data to do it.
Introducing DigiKam. This program is an opensource beast of a photo tool. Built to handle large libraries with a simple (I didn't say modern) interface and the ability to ingest any photo you give it, makes it my weapon of choice.
I do however want to give a shout out to PhotoPrism, if they had face detection and recognition released already, I probably would have settles with them first.
Step 5: Teaching Sand to see
Out of the box Digikam takes a little bit of configuration. I would suggest sticking with the default SQLite databases in most cases. Just be aware that if you are dealing with an especially large Hoard, this will make face recognition slower as the database gets larger. One of the startup questions is regarding Metadata write back, I highly suggest you enable this feature as this will allow you to more easily move out of Digikam and into a different platform more easily. (Like PhotoPrism) That being said, test with a sample first.
Once Digikam is configured, go to the "People" tab. Digikam will need to run an initial scan to find all of the faces in your Hoard, this can and will take time. I have found that the default sensitivity is pretty good to start, but I recommend enabling "YOLO v3 Detection" as it gives better results. "Work on All processor cores" seems to be broken at time of writing.
Let the facial detection do its thing, future scans are significantly faster as previous results are merged together resulting in a lower workload the more it is run. If you feel the need, you can force a full scan again.
Step 6: Finding Grandma
Once Digikam finishes finding faces, you will need to manually tell it who some of the faces are. The more photos you tag initially, the larger the starting data set will be and the higher accuracy the Individual Person recognition will be. (I had a lamp that it insisted was my grandfather at initial stages. Really odd but made me laugh way harder than I should have.)
After the first run, you will need to confirm the photos it thinks is the person you are working on. This step will be repeated over and over until a large enough sample is collected that you will eventually just be bulk confirming photos.
I encourage you to play with the sensitivity values while doing this as it can help very quickly. Running bulk face detection, people detection and confirming is the vast majority of the labor needed in this process.
Step 7: Profit
After confirming the photos to a level you are happy with, you are left with a date sorted, person tagged hoard of photos which can easily be searched in normal file viewers and is modular enough to be moved between system without having to worry about software dependencies, proprietary formats or data loss from a company you trusted to keep your hoard safe.
Best of all, the management overhead on this is pretty low and gets easier every time you do it.