Experimental browser for the Atmosphere
Towards Understanding Camera Motions in Any Video Zhiqiu Lin et 14 al. tl;dr: camera movement as binary classification and VQA: dataset and new task for VLMs. MegaSAM and MASt3R rule, unless you finetune VLM. All are bad actually. arxiv.org/abs/2504.15376
Apr 25, 2025, 8:36 AM
{
"text": "Towards Understanding Camera Motions in Any Video\n\nZhiqiu Lin et 14 al. \n\ntl;dr: camera movement as binary classification and VQA: dataset and new task for VLMs.\nMegaSAM and MASt3R rule, unless you finetune VLM. All are bad actually. \n\narxiv.org/abs/2504.15376",
"$type": "app.bsky.feed.post",
"embed": {
"$type": "app.bsky.embed.images",
"images": [
{
"alt": "",
"image": {
"$type": "blob",
"ref": {
"$link": "bafkreiguscbyl5oqdwut342rjp3fepwnondgvozfjezanqs7m3sedgkzme"
},
"mimeType": "image/jpeg",
"size": 429461
},
"aspectRatio": {
"width": 1046,
"height": 1402
}
},
{
"alt": "",
"image": {
"$type": "blob",
"ref": {
"$link": "bafkreig2rfqyjf3yak7fewczddk7vtiojkbpf6uzmwmyatl24zh7ycrmq4"
},
"mimeType": "image/jpeg",
"size": 548875
},
"aspectRatio": {
"width": 2000,
"height": 1492
}
},
{
"alt": "",
"image": {
"$type": "blob",
"ref": {
"$link": "bafkreigbxf553gdzclc2pm3wyrgedxrta4eowxbbxyy6peppd2jyv3fuva"
},
"mimeType": "image/jpeg",
"size": 603033
},
"aspectRatio": {
"width": 1026,
"height": 992
}
},
{
"alt": "",
"image": {
"$type": "blob",
"ref": {
"$link": "bafkreickzbwk6op2ctkcwihpxpwyegpfyxoalaqi5pwp55j2sda5hjdce4"
},
"mimeType": "image/jpeg",
"size": 697680
},
"aspectRatio": {
"width": 1428,
"height": 1478
}
}
]
},
"langs": [
"en"
],
"facets": [
{
"index": {
"byteEnd": 260,
"byteStart": 236
},
"features": [
{
"uri": "https://arxiv.org/abs/2504.15376",
"$type": "app.bsky.richtext.facet#link"
}
]
}
],
"createdAt": "2025-04-25T08:36:35.775Z"
}