Video dataSet Expert for machine learning

I'm seeking a dataset expert for machine learning.
This is non-skilled project.
Only need someone with highly-detailed and hard
ID: 10XXXX
First number:
1: positive
2: negative
3: soft negative
Second number:
0: DTL
1: FO
Remaining numbers:
XXXX: unique id by each category(positive, negative, soft negative)

positive: correct motions, the golfer hits the ball (full swing) at strong/medium/weak power.
Hard negative: incorrect motions, motions that look like a swing but are not (waggles, rehearsals, half moves, club pick-up/put-down)
Soft negative: general activity near the ball (walking, setting up, camera shake. Include putts here.

each video clip condition: exactly 2.0 second, 30 FPS, full body visible
Positive clip: only one swing per clip; the hitting of the ball must be centered in the 2.0s clip
Hard negative clip: 2.0s of swing-like motion with no ball strike
Soft negative clip: 2.0s snippets of walking, setup without takeway, camera shake, general movement; chips allowed; do not include putts

Include one JSON file per video clip
JSON file structure:
{
"id": "10XXXX",
"file": "FO(1,2,3)1XXXX.mp4" or DTL(1,2,3)0XXXX.mp4
"label": "swing" or "no_swing",
"view": "FO" or "DTL"
"handedness": "R" or "L"
}

file naming example: "DTL_100002.mp4" "DTL_100002.json" or "FO_110001.mp4" "FO_110001.json"

Budget: $0.25 per video clip
Total video clips: 6000 clips
Total budget: $1500

Deadline: First of all, complete 200 video clips within 3 days
80 positive: 40 DTL, 40 FO
60 hard negative: 30 DTL, 30 FO
60 soft negative: any

Complete more video clips as many as possible.