For two days, citizens poured into Belgrade for the largest protest in modern Serbian history. This occurred despite authorities' efforts to obstruct the demonstrations by halting public transportation.
Thousands of students walked into the capital, spreading messages of solidarity through smaller towns along the way. The city's streets were packed, with people occupying several key locations.
"I came for my child, for my son, so that his future can be better," a young man told DW.
Police estimated a peak turnout of 107,000. Arhiv javnih skupova (Archive of Public Gatherings), an NGO which tracks mass gatherings, reported between 275,000 and 325,000 demonstrators — possibly more.
...
Panic while honoring Novi Sad victims
The most alarming moment occurred during a 15-minute silence to honor the victims of the station collapse. A loud, unexpected noise described by witnesses as resembling a projectile or crashing aircraft, caused panic and triggered a brief stampede. Videos on social media captured the crowd scattering in fear.
Dušan Simin, who was among the crowd, told DW that it "sounded like a plane was landing from the direction of the Presidency building."
"We couldn't run away from it — we didn’t know what to do. You don’t know if something will fall on your head or hit you from the side," Simin said.
"People must have instinctively thought something was coming down the street, so they started running to the side, and we fell over each other. My wife hit her head on a lamppost. I watched her, but I couldn’t help. We still feel uneasy."
He added that they planned to seek medical attention and that the incident has already been reported to the Belgrade Center for Human Rights, which has called on citizens to reach out if they need free legal assistance.
"We will seek justice because what they did is not normal," Simin said.
Balkan news broadcaster N1 quoted military analyst Aleksandar Radic, who suggested an acoustic weapon, specifically a "sonic cannon" reportedly available to Serbian security forces, caused the sound. An opposition lawmaker echoed this claim, but police swiftly denied deploying any such device.
Image recognition depends on the amount of resources you can offer for your system. There are traditional methods of feature extractions like edge detection, histogram of oriented gradients and viola-jones, but the best performers are all convolutional neural networks.
While the term can be up for debate, you cannot separate these cases and things like LLMs and image generators, they are the same field. Generative models try to capture the distribution of the data, whereas discriminitive models try to capture the distribution of labels given the data. Unlike traditional programming, you do not directly encode a sequence of steps that manipulate data into what you want as a result, but instead you try to recover the distributions based on the data you have, and then you use the model you have made in new situations.
And generative and discriminative/diagnostic paradigms are not mutually exclusive either, one is often used to improve the other.
I understand that people are angry with the aggressive marketing and find that LLMs and image generators do not remotely live up to the hype (I myself don't use them), but extending that feeling to the entire field to the point where people say that they "loathe machine learning" (which as a sentence makes as much sense as saying that you loathe the euclidean algorithm) is unjustified, just like limiting the term AI to a single digit use cases of an entire family of solutions.