RepLiteDet-S0: a compact FTC 2026-2027 BioBuzz Pollen Detector Model for the Limelight 3A

A drop-in 8-bit TFLite model for FTC teams working in season 2026-2027 BioBuzz to detect the game element Pollen. Built by FTC 8565. The complete training scripts follow in the next release.

Backbone -> FPN-Lite fusion -> shared SSD head -> decode + NMS. The diagram shows the training graph (12,750 anchors); the exported detector drops the P5/P6 heads and runs P3 + P4 = 12,000 anchors.

Brief Comparison to the LimeLight3A Default Model

	RepLiteDet-S0 (ours)	SSD-MNV2, our pipeline	SSD-MNV2, Limelight online trainer
mAP@0.5	0.9365	0.8516	0.4605
mAP@0.75	0.7581	0.5746	0.3044
recall <10 px	0.8213	0.6823	0.3285
false positives	410	845	863
size	0.64 MB	4.96 MB	4.96 MB
latency, on a Limelight 3A	52.8 ms	86.5 ms	86.5 ms*

+10.0% mAP@0.5 and +20.4% recall on objects under 10 px than the same network trained the conventional way, at 0.64 MB against 4.96 MB and 39% less latency measured on a Limelight 3A. Against the model the standard cloud-training route produces from the same dataset at the same budget: +103% mAP@0.5 and 2.5x the sub-10 px recall. Ours runs at 320x320 while both baselines run at 300x300, so it is doing more work per frame, not less.

The weak axis is box tightness on large close-up objects; the numbers for that are below, not hidden.

Side-by-side, on real frames

Read the boxes like this:

• Green -- a detection that matches a real object (IoU >= 0.5). This is a hit.
• Yellow -- a real object the model missed: it is there in the ground truth and the model did not find it at its own threshold.
• Red -- a false positive: a box that matches no real object, either because it is on the wrong thing entirely or because it is on the right thing but positioned or sized badly enough to fail the IoU test.

Ground-truth boxes are not drawn -- a fourth box per object made the frames unreadable. Nothing is hidden: misses show yellow, unmatched detections show red, and each tile prints how many of the frame's objects that model found, at its own best-F1 threshold.

These frames are selected, and the rule is printed on the figure: every frame where we find at least 2 more objects than the retrained baseline does. Across the four, 19 of 20 found by us, 9 by the baseline, 1 by the online trainer. It shows what the difference looks like when it appears -- the tables above are what it is worth on average. Same three models, each at its own threshold.

Deploying to a Limelight 3A

Upload the 8-bit .tflite and the labels file through the Limelight web interface and select the CPU runtime, per Limelight's own detector documentation. Both files are required; a model without its labels file will not run.

What these numbers mean

The camera feeds a 320x320 image. A ball across the field is under 10 px wide; one at intake range is 50 px or more. Every metric below is measured on 3,546 real objects.

"Overlap" always means shared area divided by combined area (IoU). A box twice the size of the ball scores 0.5 even though it contains all of it -- too big is penalised exactly as hard as too small.

metric	what it measures	on the robot
mAP@0.5	did it find the ball, with a box overlapping the real one by at least 50%	"there is a ball over there" -- enough to drive toward it
mAP@0.75	same, at a 75% bar -- a loose box now counts as a miss	enough to line up an intake, not just point the robot
mAP@[.5:.95]	the average across ten strictness levels from 50% to 95%	one number for "finds it AND boxes it well"
mean best IoU	for each real ball, the overlap of the best box we drew -- averaged over every ball, so one we never found counts as ~0	box size is how you estimate distance; a loose box is a wrong distance
TP / FP	correct detections / boxes matching no real ball	FP is the robot reacting to something that is not there
best-F1	the balance of "found most of them" against "was not wrong often"	the single number if you only get one
own threshold	the confidence cutoff at which that model scores best	each model has a different one; comparing two models at the same cutoff is unfair
NMS	non-maximum suppression: the model scores thousands of candidate boxes and a real object lights up several neighbours at once, so NMS keeps the highest-scoring box and deletes anything overlapping it by more than a threshold	too permissive stacks three boxes on one ball; too aggressive merges two balls into one. Ours ships at 0.35
recall	of the balls that really are there, the fraction the model finds. Says nothing about box quality -- a box that scrapes the 50% bar counts the same as a perfect one	a miss is a ball your robot never knows about
recall by band	the same number, computed separately for each object-size range	size is distance: <10 px is the far half of the field, >45 px is intake range

Worked example: the two <10 px numbers

The eval pool holds 554 balls under 10 px -- the far half of the field. Two numbers describe how we do on them, and they answer different questions.

recall 0.8213 = how many we found. Of those 554 balls, the model produced a box that counted as a hit on 455, and missed 99. It says nothing about box quality: a box that scrapes past the 50% bar counts the same as a perfect one. This is the number that decides whether the robot knows a far ball exists at all.

mean best IoU 0.6669 = how well we boxed them, averaged over all 554. For each ball we take the best box we drew and measure the overlap, then average -- including the 99 we never found, which contribute about 0. So this one number blends finding and boxing: the misses drag it down. Back out the 82% we found and the boxes on those sit near 0.81.

Read together: we find 4 far balls in 5, and box the ones we find well. The same pair for the retrained baseline is 378 of 554 found (recall 0.6823) at mean IoU 0.5716 -- and backing out its recall the same way gives ~0.84 on what it finds. Almost the whole gap in that band is finding the ball, not boxing it.

So <10 px recall 0.8213 vs 0.3285 means: of the balls far enough away to be under 10 pixels, we find about 4 in 5 and the cloud-trained model finds about 1 in 3. In a match, that is seeing the far side of the field during auto instead of driving in blind.

What you get

file	what it is	release road map
replitedet_s0_int8.tflite	the model -- 0.64 MB, full-INT8, uploads straight to a Limelight 3A	available today
pollen_labels.txt	the labels file the device requires beside it	available today
README.md	every number here, and how it was measured	available today
ckpt-* / checkpoint	the float checkpoint, so you can retrain on YOUR data	available today
*_vp2.config	the exact training recipe we used -- nothing held back	available today
training scripts and pipeline	our training pipeline, with the tuned learning-rate schedule	releasing later

Design

Built backwards from one target: an FTC robot, watching a field, through a Limelight 3A.

• 43.9% of training objects are under 18 px, so the neck fuses high-resolution early features with deeper ones. That is where the <10 px numbers come from.
• The device runs full-integer INT8 only. No float, float16 or weight-only fallback.
• The head is a stock SSD head with in-graph TFLite_Detection_PostProcess, so the export is drop-in: upload the tflite and labels, pick the CPU runtime. A more modern head would score better and would not be drop-in.
• 0.86 M parameters. COCO pre-training is done on our architecture, class-agnostically (every annotation -> one object class), so we keep the representation and carry none of the 80-class head width: 0.64 MB vs 5.20 MB.

Where this model stands

The baseline is the network the Limelight training route produces -- plain SSD MobileNetV2 300x300, alpha 1.0 -- retrained by us, from its own COCO checkpoint, on our data, at the same 24,000 steps, with a correctly batch-scaled learning rate and its exported NMS threshold swept the same way ours was. The third comparison is that same route's own trainer, run on the same data at the same budget.

Why the baseline is a legitimate comparison: it is the same network, verified

Same architecture, our rebuild against LimeLight Trainer download:

	our retrained baseline	the Limelight-trained artifact
file size	5,202,224 bytes	5,202,224 bytes
CONV_2D / DEPTHWISE_CONV_2D	55 / 17	55 / 17
ADD / RESHAPE / CONCATENATION	10 / 13 / 2	10 / 13 / 2
QUANTIZE / DEQUANTIZE / LOGISTIC	1 / 2 / 1	1 / 2 / 1
TFLite_Detection_PostProcess	1	1

Byte-identical size, identical operator histogram: plain SSD MobileNetV2 300x300, alpha 1.0, six levels, 1,917 anchors, 1x1 convolutional box predictor. Architecture verified; their training recipe is not visible to us, which is what the three models below separate.

Separating the architecture from the training pipeline

A better network, or the same network trained better? Three models, one variable each.

model	architecture	trained by	batch	steps
RepLiteDet-S0 (this release)	RepLiteDet-S0	our pipeline	8	12k / 24k
SSD-MobileNetV2 300	SSD-MobileNetV2 300	our pipeline	8	12k / 24k
SSD-MobileNetV2 300	SSD-MobileNetV2 300	the Limelight online trainer	16	12k / 24k

• RepLiteDet-S0 vs SSD-MobileNetV2 in our pipeline isolates the architecture.
• SSD-MobileNetV2 in our pipeline vs the same network from the Limelight online trainer isolates the training pipeline, confounded with batch 16 vs 8 -- which favours the online trainer, since at equal steps batch 16 sees twice the images.

SSD-MobileNetV2 300, trained two ways. Same network, same data, 24,000 steps both sides (the online-trainer run is job eb8961b1, its log recording Steps: 24000, Batch size: 16), each read at its own best-F1 threshold:

The online-trainer run as configured: 24,000 steps, target Coral / CPU, variant Default -- we changed nothing but the step count, which is the point. 141 minutes on their H100. The trainer's own advice is on the same screen: 4,000 steps for testing, 20,000 for robot use.

	mAP@0.5	mAP@0.75	mAP@[.5:.95]	best-F1	mean IoU	TP	FP
trained by our pipeline	0.8516	0.5746	0.5426	0.8531	0.7849	3266	845
trained by the Limelight online trainer	0.4605	0.3044	0.2875	0.5092	0.5244	1506	863

+85% mAP@0.5 from the training pipeline alone, with the confound against us: batch 16 sees twice the images ours does at equal steps. The online-trainer model finds 1,506 of 3,546 objects to our 3,266.

Neither pipeline gains from the second 12k. The online trainer at 12k -> 24k: mAP@0.5 0.4509 -> 0.4605 but best-F1 0.5152 -> 0.5092 and TP 1758 -> 1506; it becomes more conservative, not better. One axis it wins outright: 12 spurious detections on 118 empty frames, against RepLiteDet-S0's 17 and 90 for the same network in our pipeline. It hallucinates least because it detects least.

By source -- the two that carry the weight

Per source, not averaged: pollen_clean is small-object-heavy, clement_pdp is the easier one, and a mean across them would read as uncertainty when it is really the gap between two distributions. pollenv1 (153 GT, 4.3%, our own smoke-test set) is reported further down and included in the pooled numbers, but is too small to head a table.

source	GT	ours mAP@0.5	ours mAP95	baseline mAP@0.5	baseline mAP95
clement_pdp	1131	0.9910	0.8482	0.9578	0.7552
pollen_clean	2262	0.8985	0.5579	0.7961	0.4408

Full pool (all 3,546 GT), each model at its OWN best-F1 threshold

model	own thr (n=1)	mAP@0.5	mAP@[.5:.95]	best-F1	mean IoU	FP
RepLiteDet-S0 (ours)	0.2	0.9365	0.6694	0.9029	0.8314	410
SSD-MobileNetV2 300 (baseline)	0.05	0.8516	0.5426	0.8531	0.7849	845
SSD-MobileNetV2 300, Limelight online trainer	0.2	0.4605	0.2875	0.5092	0.5244	863

By object size -- the reason this model exists

Size is distance: <10 px is the far half of the field, >45 px is intake range.

Recall -- the fraction of objects in that band the model finds.

band	GT	ours	SSD-MNV2, our pipeline	SSD-MNV2, LL online trainer
<10	554	0.8213	0.6823	0.3285
10-18	1001	0.9850	0.9351	0.3487
18-28	495	0.9960	0.9758	0.8121
28-45	654	0.9908	0.9786	0.7003
>45	842	0.9846	0.9857	0.9276

Mean best IoU -- how well the box lines up, averaged over every object in the band (a missed one counts ~0, so this blends finding with boxing).

band	ours	SSD-MNV2, our pipeline	SSD-MNV2, LL online trainer
<10	0.6669	0.5716	0.3209
10-18	0.8164	0.7536	0.2575
18-28	0.9009	0.8549	0.6858
28-45	0.9076	0.8752	0.6054
>45	0.8573	0.8512	0.8176

We lead every band on both channels except >45 px recall, where the baseline is ahead by 0.0011 -- inside noise, and the band we call out as our open problem.

Per source, including the one we exclude from the headline

source	GT	RepLiteDet-S0	SSD-MNV2, our pipeline	SSD-MNV2, LL online trainer
clement_pdp	1131	0.9910	0.9578	0.5389
pollen_clean	2262	0.8985	0.7961	0.4289
pollenv1 (smoke-test set, 4.3% of GT)	153	0.9894	0.9034	0.4823

pollenv1 is an early set we collected ourselves to smoke-test the pipeline. At 153 of 3,546 GT (4.3%) it is too small to support a claim on its own, so we report it and keep it out of the headline -- but it stays in the pooled numbers.

Context: how this compares to YOLO

Context, not a claim. At float, YOLO is the stronger detector -- 0.85 mAP@[.5:.95] against our 0.67 -- and that is the bar we aim at. It is listed because it is the obvious thing a team reaches for.

model (same data, same 24k budget, same eval harness)	mAP@0.5	mAP@0.75	mAP@[.5:.95]
yolov8n, float	0.9937	0.9523	0.8465
yolo11n, float	0.9926	0.9521	0.8455
yolov8n, INT8 TFLite	0.8640	0.5422	0.5158
yolo11n, INT8 TFLite	0.8396	0.4672	0.4699
this model, INT8 TFLite	0.9365	0.7581	0.6694

A Limelight 3A cannot run it. The device accepts only full-integer INT8 with a uint8 NHWC input -- float32, float16 and weight-only (w8a32/w8a16) are rejected at load. A stock YOLO export also emits one [1, 4+nc, N] tensor where the device needs 3-4 separate float32 outputs; the INT8 rows above required a custom export wrapper we wrote.

Through the one mode the device permits, YOLO loses 0.33-0.38 mAP@[.5:.95], and the loss is box tightness, not detection -- same objects found, looser boxes, confidences compressed into ~[0, 0.5]. Independently reported: YOLOv8 n/s/m/l/x lose 6.1-7.5 mAP points and YOLO11 n/s/m/l/x 6.6-8.7 under LiteRT full-integer INT8, against 0.0-0.2 for dynamic-range:

Ozturk et al., Benchmarking Quantized YOLO Models for Real-Time Object Detection on Edge Devices, arXiv:2511.13453. https://arxiv.org/pdf/2511.13453

Same weights, four export widths, identical calibration:

yolov8n, same weights	mAP@0.5	mAP@0.75	box precision @320px
float32	0.9937	0.9523	0.0000018 px
int8 weights, float activations (w8a32)	0.9940	0.9189	0.0000003 px
int8 weights, int16 activations (w8a16)	0.9936	0.9329	0.0114 px
full-integer int8 (the only mode this device runs)	0.8640	0.5422	1.584 px

The last column is the mechanism. YOLO packs box coordinates and class scores into one output tensor sharing a single 8-bit scale; at full-integer INT8 that scale is worth 1.58 px of box resolution at 320 px input, on a dataset whose median object is under 18 px. It also caps every confidence at 0.4998. Widen the activations and the damage is gone -- but those are the modes the device rejects.

Two limits on these rows: (1) it is a property of the TFLite full-integer export path, not of YOLO -- the same architectures quantize near-losslessly on Hailo (-1.6%), TensorRT (-0.05) and OpenVINO (-2.7%); (2) known mitigations exist (splitting the head's outputs, excluding the score sigmoid) and we did not implement them, so a better INT8 YOLO than this one certainly exists.

Latency

Measured on a Limelight 3A -- one-minute running average, on the device:

model	input	on device	size
RepLiteDet-S0 (ours)	320x320	52.8 ms	0.64 MB
SSD-MobileNetV2 300	300x300	86.5 ms	4.96 MB

39% faster on the device, while running a larger input -- 320x320 against their 300x300, so it is doing more work per frame, not less.

The two SSD-MobileNetV2 rows in the headline table are the same network; the device figure was measured once for that architecture and applies to both.

False positives, broken down -- read this before reacting to the raw count

410 false positives across 3546 objects, at its own best-F1 threshold. The three kinds behave very differently on a robot:

kind	count	share	what it actually is
localization	319	78%	found a REAL object, but the box is too loose to count as a hit (IoU < 0.5)
background	91	22%	a detection where there is no object -- a genuine hallucination
duplicate	0	0%	a second box on an object already detected

Duplicates are eliminated by the exported NMS threshold, not by training. Hallucinations are the smallest category:

Spurious detections on empty frames

Images containing no objects at all, which isolates hallucination:

model	false positives	per empty image
this model	17	0.1441
SSD-MobileNetV2 300, our pipeline	90	0.7627
SSD-MNV2 300, Limelight online trainer	12	0.1017

Fewer than either comparison model. This is the number that predicts how often a robot acts on nothing.

Localization FP is the open problem

Most of our false positives are loose boxes on real objects, and that share has not improved across milestones while duplicates went to zero and background FP fell. Same deficit shows up as mAP@0.75 and as the weakness on very large close-up objects.

Cause is architectural, not a training shortfall: the anchor ladder tops out near 181 px, so larger objects have no anchor that fits. Fixing the ladder is the next release's main objective.

The data this was trained and measured on

Not published. The distribution is, so you can judge transfer to your field and camera.

split	images	ground-truth boxes	negative images
train	1,931	4,352	289
valid	670	1,953	70
test	618	1,593	59
evaluation pool (valid+test)	1,288	3,546	129

Every number in this card is measured on pooled valid+test. Size = shorter box side at a 320x320 reference:

size band	GT boxes in eval	share
<10 px	554	15.6%
10-18 px	1001	28.2%
18-28 px	495	14.0%
28-45 px	654	18.4%
>45 px	842	23.7%

Training set: mean 39.3 px, median 23.4 px, 43.9% under 18 px -- which is why this card reports recall per band rather than pooled.

source	eval GT	what it is
clement_pdp	1,131	the easier source: larger objects, median ~31 px
pollen_clean	2,262	the hard one: median ~13 px, ~26% under 10 px
pollenv1	153	an early set we collected ourselves to smoke-test the pipeline

Leakage check. Video-derived, so split in chunks with guard frames dropped at boundaries: 0 of 670 valid and 0 of 618 test images collide with training at a perceptual-hash distance <= 3.

Limits

• Latency is a one-minute running average measured on a Limelight 3A, taken with the vision pipeline running as it normally would -- what a robot actually sees, rather than an isolated benchmark figure.

• Training data not published. Reproducible in shape, not in absolute value.

• Single class. Multi-class retraining needs num_classes changed and a re-export.

• We slightly lose on very large, close-up objects. >45 px recall 0.9846 vs the baseline's 0.9857, that is the only band it leads. Its ladder runs six levels to 1x1; ours tops out near 181 px, so an object filling the view has no anchor that fits. Fixing in next release.

• Every comparison here is 24,000 training steps on the same images, each model read at its own best-F1 threshold. The Limelight online trainer runs at batch 16 against our 8, which favours it.

• Post-processing was tuned, and the tuning costs band recall. The exported NMS IoU is 0.35, swept on the eval pool. Against 0.45 it is better on everything pooled -- mAP@0.5 0.9365 vs 0.9311, best-F1 0.9029 vs 0.8837, FP 410 vs 508 -- and it pays for that in two size bands, <10px recall 0.8213 vs 0.8321 and >45px 0.9846 vs 0.9964, because pollen occurs in crowded clusters that tighter suppression merges. If your scene is cluster-heavy, 0.45 is a one-flag re-export and no retraining -- ask us, or do it yourself once the export path ships.

Where this goes

For FTC 2026-2027 BioBuzz game element Pollen on a Limelight 3A: small enough to leave headroom, faster than the alternatives measured here, more accurate than the same network trained conventionally, and drop-in for the detector pipeline you already have. Box tightness on large objects is the open axis and the next release's target.

Built to be retuned, and the checkpoint plus training pipeline are the next thing we ship. If you run this one on your own hardware, we would like to hear what you measure.

Retraining on your own data

The float checkpoint and the exact config are here. The training pipeline we used is not yet -- it follows in the next release. What that means practically:

This is not a stock TFOD architecture. The config declares a feature extractor ssd_replitedet_s0_fpn_keras, and the anchor generator is selected by the _vp2 filename suffix. Neither exists in a stock Object Detection API install, so model_builder.build on this config fails with an unknown-feature-extractor error until our code is on the path.

If you would rather not wait, everything you need to write your own trainer is published: the config carries the full recipe, the checkpoint carries the layer names and shapes, and the anchor geometry -- the one thing NOT recoverable from either -- is written out below.

The anchor geometry, in full

The config's multiscale_anchor_generator block declares 6 anchors per location, but our generator overrides where they sit. Build stock anchors from that proto and you get a silently worse model, because the head weights were trained against these:

level	stride	grid	6 anchors per cell, as (size px @ offset within the cell)
P3	8	40x40	16 @ (2,2), 16 @ (6,2), 16 @ (2,6), 16 @ (6,6), 32 @ (4,4), 45 @ (4,4)
P4	16	20x20	64, 90.5, 90.5, 128, 128, 181 -- all square, all centred at (8,8)
P5	32	10x10	128, 181, 181, 256, 256, 362 -- centred (head dropped at export)
P6	64	5x5	256, 362, 362, 512, 512, 724 -- centred (head dropped at export)

Sizes are square, in pixels at the 320x320 input. P4-P6 are base x {1, sqrt2, 2} x {1, sqrt2} with bases 64 / 128 / 256, which is why 90.5 and 128 each appear twice. The four shifted 16 px anchors at P3 are the virtual P2: they give sub-cell coverage at stride 8 without paying for a stride-4 level.

The shipped detector keeps P3 + P4 only -- 9,600 + 2,400 = 12,000 anchors. The P5 and P6 heads are dropped at export; the training graph has all four levels and 12,750.

Two things that will cost you a day if you miss them:

1. Keep the config filename ending in _vp2.config. Once our pipeline is out, the anchor generator is selected by filename. Rename it and you train stock anchors against a vP2 checkpoint, with no error message and a large accuracy loss.
2. fine_tune_checkpoint_type: full loads the detection heads too. detection loads only the backbone and leaves the heads random -- correct if you change the number of classes, wrong if you do not.

License

Apache License 2.0 -- see LICENSE. This covers the model weights, the TFLite export and the training config shipped here. It does not cover the training data, which is not published, nor the third-party baseline model referenced for comparison, which is not redistributed here.

Models produced by the Limelight online trainer are not redistributed here. Train your own if you want to reproduce that row.