I've been watching these matches that have been judged by DeepStrike, an artificial intelligence boxing scoring system produced by Jabbr. I have been going out of my way to find mistakes with DeepStrike, and 99 percent of the time, I'm wrong. DeepStrike is incredibly accurate.
Jabbr who makes DeepStrike has been uploading some recent fights to show how accurate it is and how it would help clean up the sport of boxing. We all know that Compubox on the best of days is only moderately accurate. Too many times there are huge discrepancies in punches landed, and in those instances, it is suspiciously favorable to the A-side fighter, practically exclusively favorable.
There are also the issues the sport has with poor judging, and it appears that DeepStrike offers the public the opportunity to help clean up the sport and remove human error, whether it's from incompetence, bias or corruption. Of course with the issue of corruption, there will be a lot of resistance toward replacing human judges with AI. These promotional companies and networks are not going to willingly lose the ability to influence the outcome of a match, so I'm not confident that DeepStrike's services will be enlisted anytime soon.
Nonetheless, I believe that we, as boxing fans, should support Jabbr and Deepstrike. I think we should push to have DeepStrike the standard for judging professional boxing matches. As I understand it, although DeepStrike cannot visual represent the fight in real time (being 3x slower than the actual fight), it can determine pouches landed and missed in real time. So at the moment, it is superior to Compubox.
As an example of DeepStrikes capabilities, we have Canelo vs Golovkin II. I encourage you to try to find something the AI missed. I haven't finished the fight yet, but so far, every time I have though the AI missed or misrepresented a strike, I saw I was the one who was incorrect after slowing the fight down and watching in slow-motion.
Golovkin vs Canelo II:
https://www.youtube.com/watch?v=LoLieTFrw5w&t=694s
(BTW: Let's see how long it takes aboutfkntime to show his ugly mug in this thread. I give it less than 24 hours :lol1:)
I've been watching these matches that have been judged by DeepStrike, an artificial intelligence boxing scoring system produced by Jabbr. I have been going out of my way to find mistakes with DeepStrike, and 99 percent of the time, I'm wrong. DeepStrike is incredibly accurate.
Jabbr who makes DeepStrike has been uploading some recent fights to show how accurate it is and how it would help clean up the sport of boxing. We all know that Compubox on the best of days is only moderately accurate. Too many times there are huge discrepancies in punches landed, and in those instances, it is suspiciously favorable to the A-side fighter, practically exclusively favorable.
There are also the issues the sport has with poor judging, and it appears that DeepStrike offers the public the opportunity to help clean up the sport and remove human error, whether it's from incompetence, bias or corruption. Of course with the issue of corruption, there will be a lot of resistance toward replacing human judges with AI. These promotional companies and networks are not going to willingly lose the ability to influence the outcome of a match, so I'm not confident that DeepStrike's services will be enlisted anytime soon.
Nonetheless, I believe that we, as boxing fans, should support Jabbr and Deepstrike. I think we should push to have DeepStrike the standard for judging professional boxing matches. As I understand it, although DeepStrike cannot visual represent the fight in real time (being 3x slower than the actual fight), it can determine pouches landed and missed in real time. So at the moment, it is superior to Compubox.
As an example of DeepStrikes capabilities, we have Canelo vs Golovkin II. I encourage you to try to find something the AI missed. I haven't finished the fight yet, but so far, every time I have though the AI missed or misrepresented a strike, I saw I was the one who was incorrect after slowing the fight down and watching in slow-motion.
Golovkin vs Canelo II:
https://www.youtube.com/watch?v=LoLieTFrw5w&t=694s
(BTW: Let's see how long it takes aboutfkntime to show his ugly mug in this thread. I give it less than 24 hours :lol1:)
what a total muppet
typical casual fan stupidity
there is literally no reason for a boxing fan to ever mention compubox..... because, ' punches landed ' is NOT official scoring criteria
so..... wanking on about a better version of compubox, simply because you have no idea who won the fight..... is just casual fan stupidity
I will explain how this sport works.....
1) ' punches landed ' is not official scoring criteria, you muppet..... " clean hard punches " is the official scoring criteria that has you so confused..... which is why Olympic-level amateurs often need to adapt their style when they turn pro, because professional boxing is not about scoring points..... pro boxing is " The Hurt Game " son, and don't you ever forget it..... " clean, hard, punches ", nothing else
not probing or pushed jabs, not ineffective glancing hooks / uppercuts, not meaningless taps to the body, nothing that hits gloves forearms shoulders or the back of the head..... NONE OF THAT SHlT, is official scoring criteria..... judges will not score anything that does not land clean, or anything that is not thrown with bad intentions
back in the day compubox was regarded as a total joke..... they would often score 40+ punches for a fighter who landed about 15..... crazy high punch stats were commonplace because compubox counted EVERYTHING, including ineffective punches that did not land cleanly like the examples I gave above..... compubox is obviously not as bad as it was 30 years ago, but it is still nowhere near accurate enough, or capable enough (see point 2 below), to judge a professional boxing match
2) " clean hard punches "..... is only 1 of the 4 official scoring criteria..... the other three being - Defense / Ring Generalship / Effective Aggression..... but unfortunately, compubox - and you casual fan idiots - know nothing about 3/4 of the official scoring criteria
thinking that two guys with a clicker can replace a bench of trained judges is fkn ridiculous
not understanding the official scoring criteria is probably why you casuals cry robbery all the time and never seem to know who won the fight
FACT: Defense..... Ring Generalship..... and Effective Aggression..... are also official scoring criteria, you muppet..... despite the fact that compubox - and you casuals - have no idea what they are
Give me an example of a fight that DeepStrike judged, and show me how it scored Ring Generalship..... I will wait
...
I've been watching these matches that have been judged by DeepStrike, an artificial intelligence boxing scoring system produced by Jabbr. I have been going out of my way to find mistakes with DeepStrike, and 99 percent of the time, I'm wrong. DeepStrike is incredibly accurate.
Jabbr who makes DeepStrike has been uploading some recent fights to show how accurate it is and how it would help clean up the sport of boxing. We all know that Compubox on the best of days is only moderately accurate. Too many times there are huge discrepancies in punches landed, and in those instances, it is suspiciously favorable to the A-side fighter, practically exclusively favorable.
There are also the issues the sport has with poor judging, and it appears that DeepStrike offers the public the opportunity to help clean up the sport and remove human error, whether it's from incompetence, bias or corruption. Of course with the issue of corruption, there will be a lot of resistance toward replacing human judges with AI. These promotional companies and networks are not going to willingly lose the ability to influence the outcome of a match, so I'm not confident that DeepStrike's services will be enlisted anytime soon.
Nonetheless, I believe that we, as boxing fans, should support Jabbr and Deepstrike. I think we should push to have DeepStrike the standard for judging professional boxing matches. As I understand it, although DeepStrike cannot visual represent the fight in real time (being 3x slower than the actual fight), it can determine pouches landed and missed in real time. So at the moment, it is superior to Compubox.
As an example of DeepStrikes capabilities, we have Canelo vs Golovkin II. I encourage you to try to find something the AI missed. I haven't finished the fight yet, but so far, every time I have though the AI missed or misrepresented a strike, I saw I was the one who was incorrect after slowing the fight down and watching in slow-motion.
Golovkin vs Canelo II:
https://www.youtube.com/watch?v=LoLieTFrw5w&t=694s
(BTW: Let's see how long it takes aboutfkntime to show his ugly mug in this thread. I give it less than 24 hours :lol1:)
what a total muppet
typical casual fan stupidity
there is literally no reason for a boxing fan to ever mention compubox..... because, ' punches landed ' is NOT official scoring criteria
so..... wanking on about a better version of compubox, simply because you have no idea who won the fight..... is just casual fan stupidity
I will explain how this sport works.....
1) ' punches landed ' is not official scoring criteria, you muppet..... " clean hard punches " is the official scoring criteria that has you so confused..... which is why Olympic-level amateurs often need to adapt their style when they turn pro, because professional boxing is not about scoring points..... pro boxing is " The Hurt Game " son, and don't you ever forget it..... " clean, hard, punches ", nothing else
not probing or pushed jabs, not ineffective glancing hooks / uppercuts, not meaningless taps to the body, nothing that hits gloves forearms shoulders or the back of the head..... NONE OF THAT SHlT, is official scoring criteria..... judges will not score anything that does not land clean, or anything that is not thrown with bad intentions
back in the day compubox was regarded as a total joke..... they would often score 20+ punches for a fighter who landed about 7..... crazy high punch stats were commonplace because compubox counted EVERYTHING, including ineffective punches that did not land cleanly like the examples I gave above..... compubox is obviously not as bad as it was 30 years ago, but it is still nowhere near accurate enough, or capable enough (see point 2 below), to judge a professional boxing match
2) " clean hard punches "..... is only 1 of the 4 official scoring criteria..... the other three being - Defense / Ring Generalship / Effective Aggression..... but unfortunately, compubox - and you casual fan idiots - know nothing about 3/4 of the official scoring criteria
thinking that two guys with a clicker can replace a bench of trained judges is fkn ridiculous
not understanding the official scoring criteria is probably why you casuals cry robbery all the time and never seem to know who won the fight
FACT: Defense..... Ring Generalship..... and Effective Aggression..... are also official scoring criteria, you muppet..... despite the fact that compubox - and you casuals - have no idea what they are
Give me an example of a fight that DeepStrike judged, and show me how it scored Ring Generalship..... I will wait
...
DeepStrike is capable of determining the level of impact a clean punch has on a boxer. In other words, it is capable of determining the "effect" of a punch by qualifying it on a six degree scale, from no contact to max contact. In the other thread, a few forum members asked how it is capable of accomplishing this and one claimed that it is incapable of determining the strength of impact or, more specifically, the "effect" of the punch. I'm going to state what the creators of DeepStrike stated regarding its capability of assessing the level of impact a punch has on its target.
DeepStrike is capable of determining the strength of impact based on various factors from visual cues, usually acquired from numerous video feeds around the ring. From this, it is able to qualify the level of impact and give it a rating. The quality rating is a regression model that funnels various features into a single score, briefly it considers 3 things:
Cleanliness: Does the punch land cleanly, or is partially or fully blocked
Accuracy: Does the punch land in a way that can transfer power, or is it just grazing the target or being rolled away
Effect: Does the target show visible effect when hit, i.e. is the head bumping and how much is it bumping
DeepStrike is constantly being developed and improved upon, so it's an evolving system. The more it is used, the better it becomes. The creators of DeepStrike (Jabbr) welcome positive feedback in order to make the system even better, so if anyone has an idea on how to improve it or make it better than it is, or if a mistake is found, then Jabbr will welcome the feedback.
(One of my threads highlighting a single fight analysis has been closed for some unknown reason. However, I want to add this because it was something that consistently came up in the previous thread.)
Ah ha! Here we go. They did one on Duran vs Hagler. Let's see who won. Do you have your original scorecard?
https://www.youtube.com/watch?v=FMWFBkoaJVU
I rewatched this fight last night. I don't want to spoil it for anyone, but all I have to say about this fight is "Damn, Duran is hard to hit." I also have to say "Damn, Hagler was a badass inside fighter." I forgot how badass Hagler is on the inside.
I recommend comparing this to your own personal scorecard (if you ever scored or rescored the fight). Very interesting.
We should be supporting Jabbr, the company who produces DeepStrike.
(By the way, why is this not getting any responses by the Loma fight, people can't let it go?)
I'd be for anything that takes human error out of the equation to be honest. way too many garbage scorecards over the years to say the current system is the best option
Absolutely. It's been so bad lately. It's ruining the sport. From seeing these young, upcoming boxers get robbed to even the big names, I just don't have the stomach for it anymore. Most the robberies that I see occur happen on undercards (mostly on Top Rank to be honest). It turns the stomach man. Sports don't follow scripts, which is why I don't consider boxing to be much of a sport anymore.
lol I hope not. From my point of view, whatever model is used, it's based on those four criteria. In fact, it's more advanced than my own judging criteria. My criteria is that with clean punches, the whole knuckle of the glove has to land. Partially blocked shots are not considered clean punches. People disagree on that, and DEEPSTRIKE will count punches that punches through gloves or predominately land, despite catching glove. I also had three different punches. Regular punches. Hard punches, and then damaging punches (staggering punches). This AI actually assesses five different degrees of damaging punches.
What I find extremely interesting is that it graphs out and keeps track of combination punches. It's a fascinating graph. If you get a chance, check out the end of each round. It also assess who's applying pressure and who's being the aggressor.
Apparently, they've been using this for amateur bouts around the country to great success. With the right setup, it can provide a visual presentation and statistical presentation in real time. This really is the future.
And in regard to different models providing varied interpretations, I think we can all qualify what is a landed punch and what is a missed punch. It may not be 100 percent perfect, but it is clearly superior to how judging is conducted at the moment.
I'd be for anything that takes human error out of the equation to be honest. way too many garbage scorecards over the years to say the current system is the best option
With boxing, they'd come out with 3 different AI models that score 3 different ways and we'd be right back to where we are now with judges all seeing something different and scoring fights completely opposite of one another despite sitting 15 feet from one another lol
lol I hope not. From my point of view, whatever model is used, it's based on those four criteria. In fact, it's more advanced than my own judging criteria. My criteria is that with clean punches, the whole knuckle of the glove has to land. Partially blocked shots are not considered clean punches. People disagree on that, and DEEPSTRIKE will count punches that punches through gloves or predominately land, despite catching glove. I also had three different punches. Regular punches. Hard punches, and then damaging punches (staggering punches). This AI actually assesses five different degrees of damaging punches.
What I find extremely interesting is that it graphs out and keeps track of combination punches. It's a fascinating graph. If you get a chance, check out the end of each round. It also assess who's applying pressure and who's being the aggressor.
Apparently, they've been using this for amateur bouts around the country to great success. With the right setup, it can provide a visual presentation and statistical presentation in real time. This really is the future.
And in regard to different models providing varied interpretations, I think we can all qualify what is a landed punch and what is a missed punch. It may not be 100 percent perfect, but it is clearly superior to how judging is conducted at the moment.
With boxing, they'd come out with 3 different AI models that score 3 different ways and we'd be right back to where we are now with judges all seeing something different and scoring fights completely opposite of one another despite sitting 15 feet from one another lol
Let's run some classic controversial classics through and see if everyone agrees with Jabbr DEEPSTRIKE scoring. Compubox, of course, is not used to actually judge a fight. Very interesting thread.
Ah ha! Here we go. They did one on Duran vs Hagler. Let's see who won. Do you have your original scorecard?
https://www.youtube.com/watch?v=FMWFBkoaJVU
Willow The Wisp requested that we run some controversial classics. I'll look through what Jabbr has posted, but right off the bat, we have Devin Haney vs. Lomachenko. Judges had it 115-113 twice and 116-112​. I just heard on ThaBoxingVoice that Brad Goodman had initial scored the fight for Loma, but when he watched it on TV, he said Haney won 8 rounds to 4, making it 116-112 on his card. Let's see if Brad Goodman knows what he's talking about. Here's the fight scored by unbiased artificial intelligence.
https://www.youtube.com/watch?v=wqQkA3My_xY&t=27s
BTW: I home a modern controversial classic will suffice, Willow The Wisp.
Let's run some classic controversial classics through and see if everyone agrees with Jabbr DEEPSTRIKE scoring. Compubox, of course, is not used to actually judge a fight. Very interesting thread.
Will do. They did the Lomachenko vs Haney fight, which is the most recent controversial fight. Judges had it 115-113 twice and 116-112. Let's see if this was an actual robbery. (BTW: How much of a discrepancy is required to consider a fight a robbery? And should I post that fight in this thread or make a new thread for it?)
MY SCORE ORIGINALLY
GOLOVKIN
ALVAREZ
1
10
9
2
9
10
3
10
9
4
10
9
5
10
9
6
9
10
7
9
10
8
10
9
9
10
9
10
10
9
11
10
9
12
10
9
TOTAL:
117
111
NOTE: I had round 11 as a swing round. Canelo, for the first time, was up against the ropes, just as he was in the first fight. Something happened in the last round, and it's no longer the same Canelo. His defense is gone. He's getting pieced up. Canelo relied back at the end but not enough to win it.
DEEPSTRIKE'S SCORE
GOLOVKIN
ALVAREZ
1
10
9
2
9
10
3
10
9
4
10
9
5
10
9
6
9
10
7
10
9
8
10
9
9
10
9
10
10
9
11
10
9
12
10
9
TOTAL:
118
110
NOTE: Round Two I labeled as a swing round, even though Golovkin outlanded Canelo and did so with harder punches in general. However, Canelo landed the hardest punch in that round, and the total punches landed are close enough to have that as a swing round for either fighter. You can make an argument for either fighter winning that round. Canelo wins this round.
Round four Golovkin landed nearly 3x as many punches as Canelo, which leads me to ask what qualifies as a 10-8 round? What is the threshold for round dominance?
Round Six, Golovkin outlanded Canleo and landed the hardest punch in the round. However, Canelo landed more impactful punches in total, so you have to go with Canelo winning this round. He showed a little more aggression too, when landing his punches. This is the first round that you can make a good case for Canelo winning.
Round Seven, while I scored this for Canelo, DeepStrike showed that I was wrong, grossly wrong. This was a clear Golovkin round, where he landed the most impactful punches, more punches and the hardest punch.
Round Eight, Golovkin edged it out according to Deepstrike because he landed more punches (14 more punches than Alvarez), even though both landed and equal amount of hard punches. What pushes it for Golovkin according to DeepStrike is that he was the aggressor for a greater portion of the round, and he was applying more pressure (Ring Generalship). It was closer than I remember its being, however. Also, Canelo is only able to land two-punch combinations at most, while Golovkin would landed three and four-punch combinations.
Round Nine, both landed about an equal number of hard punches, with Golovkin landed the hardest punch of the round. However, Golovkin outlanded Canelo by 19 punches. He was also predominantly the effective aggressor.
Round Ten, Golovkin staggered Canelo this round, and he landed twice as many punches. DeepStrike shows Golovkin clearly won this round.
Round Eleven, first half of this round, Golovkin is kicking Canelo's ass. I think Canelo's chin is comparably to Golovkin's. This guy can take punches. Back half of the fight, Canelo lands some bombs, the hardest punches of the round. But overall, for high impact punches, Golovkin landed 8 and Canelo 6. And overall, Golovkin landed 24 punches to Canelo's 12. This is a Golovkin round.
Round Twelve, Golovkin wins the round. He landed 12 high impact punches to Canelo's 4. Golovkin also outlanded Canelo in total punches landed. He was the ring general most of the round and predominately the aggressor according to DeepStrike.
I scored the fight 117 to 111 for Golovkin. However, this fight should have been scored 118 to 110 for Golovkin, which means that this was highway robbery again.
Let's run some classic controversial classics through and see if everyone agrees with Jabbr DEEPSTRIKE scoring. Compubox, of course, is not used to actually judge a fight. Very interesting thread.
It might help clean up boxing, and that's precisely why it won't be adopted. There's a lot of things that could be done to clean up boxing, none of which require any technological adoption, etc, and none have gotten traction. Why? Because boxing is corrupt, and the powers in it prefer it to stay corrupt so they can keep making money.
Like politics! Commerce!
This can only be good news! CompuBox isn't inaccurate, it is a fake tool used to sway the viewer's opinion. Mayweather-Pacquiao is a great example of how ridiculous the fake CompuBox numbers can be. The Fury-Ngannou numbers also had Ngannou landing many punches that were cuffing and/or non-scoring. Will open up a can of worms if this DEEPSTRIKE develops, and we can compare the true numbers with the CompuBox numbers for any given fight...
It might help clean up boxing, and that's precisely why it won't be adopted. There's a lot of things that could be done to clean up boxing, none of which require any technological adoption, etc, and none have gotten traction. Why? Because boxing is corrupt, and the powers in it prefer it to stay corrupt so they can keep making money.
Unfortunately, this is the truth of the matter. But I'm not going to lose hope for the sport. We just need to have a centralized governing body so that spurious influences (e.g. Top Rank) can't influence state bodies (e.g. NAC) and corrupt the sport.
At least we could rule out promoter influence and outright bribery. Those are no small things in boxing.
I'm hoping man. I don't think there's anyway to eliminate corruption completely, but I think we can push toward a centralized governing body for the sport, a federal sporting agency to oversee the states. Atlas and others pushed for this but to no avail. I think senator McCain spearheaded this movement but unfortunately didn't live long enough to get it over the line. I don't know of anyone else in a position of power who's pushing toward cleaning up the sport.
Anyway, this would go a long way toward providing an even playing field.
It might help clean up boxing, and that's precisely why it won't be adopted. There's a lot of things that could be done to clean up boxing, none of which require any technological adoption, etc, and none have gotten traction. Why? Because boxing is corrupt, and the powers in it prefer it to stay corrupt so they can keep making money.